diff --git a/.github/actions/detect-backend-changes/action.yml b/.github/actions/detect-backend-changes/action.yml new file mode 100644 index 00000000000..af01038f294 --- /dev/null +++ b/.github/actions/detect-backend-changes/action.yml @@ -0,0 +1,48 @@ +name: "Detect backend-relevant changes" +description: >- + Classify the pull request's changed files with .circleci/scripts/classify_changes.sh + and expose decision=run|skip. decision=skip means only ui/**, **.md or **.mdx files + changed, so callers can short-circuit expensive steps while the job still completes + successfully and satisfies its required status check. The decision defaults to run for + any non pull_request event or whenever the changed set cannot be resolved, so tests are + never skipped when the classification is uncertain. + +outputs: + decision: + description: "run when backend-relevant files changed, otherwise skip" + value: ${{ steps.classify.outputs.decision }} + +runs: + using: composite + steps: + - id: classify + shell: bash + env: + BASE_SHA: ${{ github.event.pull_request.base.sha }} + run: | + set -uo pipefail + if [ -z "${BASE_SHA:-}" ]; then + echo "detect-backend-changes: not a pull_request event; running job" + echo "decision=run" >> "${GITHUB_OUTPUT}" + exit 0 + fi + if ! git fetch --no-tags --depth=1 origin "${BASE_SHA}" >/dev/null 2>&1; then + echo "detect-backend-changes: could not fetch base ${BASE_SHA}; running job" + echo "decision=run" >> "${GITHUB_OUTPUT}" + exit 0 + fi + changed="$(git diff --name-only "${BASE_SHA}" HEAD 2>/dev/null)" || { + echo "detect-backend-changes: git diff failed; running job" + echo "decision=run" >> "${GITHUB_OUTPUT}" + exit 0 + } + if [ -z "${changed}" ]; then + echo "detect-backend-changes: no changed files vs ${BASE_SHA}; skipping job" + echo "decision=skip" >> "${GITHUB_OUTPUT}" + exit 0 + fi + echo "detect-backend-changes: changed files vs ${BASE_SHA}:" + printf '%s\n' "${changed}" | sed 's/^/ /' + decision="$(printf '%s\n' "${changed}" | bash .circleci/scripts/classify_changes.sh backend)" || decision="run" + echo "detect-backend-changes: decision=${decision}" + echo "decision=${decision}" >> "${GITHUB_OUTPUT}" diff --git a/.github/workflows/_test-unit-base.yml b/.github/workflows/_test-unit-base.yml index 25c6d4a7019..9fd81b27f3b 100644 --- a/.github/workflows/_test-unit-base.yml +++ b/.github/workflows/_test-unit-base.yml @@ -45,12 +45,18 @@ jobs: name: Run tests runs-on: ubuntu-latest timeout-minutes: ${{ inputs.timeout-minutes }} + outputs: + decision: ${{ steps.changes.outputs.decision }} steps: - uses: actions/checkout@08eba0b27e820071cde6df949e0beb9ba4906955 # v4.3.0 with: persist-credentials: false + - name: Detect backend-relevant changes + id: changes + uses: ./.github/actions/detect-backend-changes + - name: Set up Python uses: actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065 # v5.6.0 with: @@ -72,16 +78,19 @@ jobs: ${{ runner.os }}-uv- - name: Install dependencies + if: steps.changes.outputs.decision != 'skip' run: | .github/scripts/uv_sync_with_retries.sh --frozen --group ci --group proxy-dev --extra google --extra proxy --extra semantic-router - name: Generate Prisma client + if: steps.changes.outputs.decision != 'skip' env: PRISMA_BINARY_CACHE_DIR: ${{ runner.temp }}/prisma-cache run: | uv run --no-sync prisma generate --schema litellm/proxy/schema.prisma - name: Run tests + if: steps.changes.outputs.decision != 'skip' env: TEST_PATH: ${{ inputs.test-path }} MAX_FAILURES: ${{ inputs.max-failures }} @@ -114,7 +123,7 @@ jobs: fi - name: Save coverage report - if: always() + if: always() && steps.changes.outputs.decision != 'skip' uses: actions/upload-artifact@4cec3d8aa04e39d1a68397de0c4cd6fb9dce8ec1 # v4.6.1 with: name: coverage-${{ inputs.artifact-name }}-${{ github.run_id }}-${{ github.run_attempt }} @@ -124,7 +133,7 @@ jobs: upload-coverage: name: Upload coverage to Codecov needs: run - if: always() + if: always() && needs.run.outputs.decision != 'skip' runs-on: ubuntu-latest permissions: contents: read diff --git a/.github/workflows/create_daily_oss_branch.yml b/.github/workflows/create_daily_oss_branch.yml new file mode 100644 index 00000000000..43de4a0e75f --- /dev/null +++ b/.github/workflows/create_daily_oss_branch.yml @@ -0,0 +1,61 @@ +name: Create Daily OSS Branch + +on: + schedule: + - cron: "0 16 * * 1-5" # 9am PT during daylight saving time, weekdays. + workflow_dispatch: + inputs: + date: + description: "Branch date in YYYY_MM_DD format. Defaults to today's UTC date." + required: false + type: string + +permissions: + contents: write + +jobs: + create-oss-branch: + if: github.repository == 'BerriAI/litellm' + runs-on: ubuntu-latest + timeout-minutes: 10 + + steps: + - name: Checkout repository + uses: actions/checkout@08eba0b27e820071cde6df949e0beb9ba4906955 # v4.3.0 + with: + fetch-depth: 0 + persist-credentials: false + + - name: Create dated OSS branch + env: + GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }} + REQUESTED_DATE: ${{ inputs.date }} + run: | + set -euo pipefail + + if [ -n "${REQUESTED_DATE}" ]; then + if ! echo "${REQUESTED_DATE}" | grep -Eq '^[0-9]{4}_[0-9]{2}_[0-9]{2}$'; then + echo "::error::date must use YYYY_MM_DD format, got '${REQUESTED_DATE}'" + exit 1 + fi + BRANCH_DATE="${REQUESTED_DATE}" + else + BRANCH_DATE="$(date -u +'%Y_%m_%d')" + fi + + BRANCH_NAME="litellm_oss_daily_${BRANCH_DATE}" + echo "Creating branch: ${BRANCH_NAME}" + + git config user.name "github-actions[bot]" + git config user.email "github-actions[bot]@users.noreply.github.com" + + git fetch origin main "${BRANCH_NAME}" || true + + if git show-ref --verify --quiet "refs/remotes/origin/${BRANCH_NAME}"; then + echo "Branch ${BRANCH_NAME} already exists. Skipping creation." + exit 0 + fi + + git checkout -b "${BRANCH_NAME}" origin/main + git push "https://x-access-token:${GITHUB_TOKEN}@github.com/${GITHUB_REPOSITORY}.git" "${BRANCH_NAME}" + echo "Successfully created and pushed branch: ${BRANCH_NAME}" diff --git a/.github/workflows/oss_daily_guardrails.yml b/.github/workflows/oss_daily_guardrails.yml new file mode 100644 index 00000000000..950c51c9b60 --- /dev/null +++ b/.github/workflows/oss_daily_guardrails.yml @@ -0,0 +1,50 @@ +name: OSS Daily Guardrails + +on: + push: + branches: + - "litellm_oss_daily_20*" + pull_request: + branches: + - "litellm_oss_daily_20*" + - litellm_internal_staging + +permissions: + contents: read + +concurrency: + group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }} + cancel-in-progress: true + +jobs: + oss-safe-checks: + name: Run OSS daily safe checks + if: startsWith(github.ref_name, 'litellm_oss_daily_20') || startsWith(github.head_ref, 'litellm_oss_daily_20') || startsWith(github.base_ref, 'litellm_oss_daily_20') + runs-on: ubuntu-latest + timeout-minutes: 10 + + steps: + - name: Checkout repository + uses: actions/checkout@08eba0b27e820071cde6df949e0beb9ba4906955 # v4.3.0 + with: + persist-credentials: false + + - name: Set up Python + uses: actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065 # v5.6.0 + with: + python-version: "3.12" + + - name: Set up uv + uses: astral-sh/setup-uv@37802adc94f370d6bfd71619e3f0bf239e1f3b78 # v7 + with: + version: "0.10.9" + + - name: Run secret scan test + run: | + uv run --frozen --with 'pytest==9.0.2' pytest tests/litellm/test_no_hardcoded_secrets.py -v + + - name: Run Ruff + run: | + uv sync --frozen + cd litellm + uv run --no-sync ruff check . diff --git a/.github/workflows/test-unit-core-utils.yml b/.github/workflows/test-unit-core-utils.yml index e563679660b..d6d6353238f 100644 --- a/.github/workflows/test-unit-core-utils.yml +++ b/.github/workflows/test-unit-core-utils.yml @@ -7,10 +7,6 @@ on: - litellm_internal_staging - litellm_oss_staging - "litellm_**" - paths-ignore: - - "ui/**" - - "**.md" - - "**.mdx" permissions: contents: read diff --git a/.github/workflows/test-unit-documentation.yml b/.github/workflows/test-unit-documentation.yml index 2c3d6e46618..03f9f0a510b 100644 --- a/.github/workflows/test-unit-documentation.yml +++ b/.github/workflows/test-unit-documentation.yml @@ -7,10 +7,6 @@ on: - litellm_internal_staging - litellm_oss_staging - "litellm_**" - paths-ignore: - - "ui/**" - - "**.md" - - "**.mdx" permissions: contents: read @@ -36,6 +32,10 @@ jobs: path: docs/my-website persist-credentials: false + - name: Detect backend-relevant changes + id: changes + uses: ./.github/actions/detect-backend-changes + - name: Set up Python uses: actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065 # v5.6.0 with: @@ -57,10 +57,12 @@ jobs: ${{ runner.os }}-uv- - name: Install dependencies + if: steps.changes.outputs.decision != 'skip' run: | .github/scripts/uv_sync_with_retries.sh --frozen --group ci --group proxy-dev --extra google --extra proxy --extra semantic-router - name: Generate Prisma client + if: steps.changes.outputs.decision != 'skip' env: PRISMA_BINARY_CACHE_DIR: ${{ runner.temp }}/prisma-cache run: | @@ -68,6 +70,7 @@ jobs: # Run the same documentation tests that CircleCI ran (as direct Python scripts) - name: Run documentation validation tests + if: steps.changes.outputs.decision != 'skip' run: | uv run --no-sync python ./tests/documentation_tests/test_env_keys.py uv run --no-sync python ./tests/documentation_tests/test_router_settings.py diff --git a/.github/workflows/test-unit-enterprise-routing.yml b/.github/workflows/test-unit-enterprise-routing.yml index 7a9b8b00f26..13136c968d1 100644 --- a/.github/workflows/test-unit-enterprise-routing.yml +++ b/.github/workflows/test-unit-enterprise-routing.yml @@ -7,10 +7,6 @@ on: - litellm_internal_staging - litellm_oss_staging - "litellm_**" - paths-ignore: - - "ui/**" - - "**.md" - - "**.mdx" permissions: contents: read diff --git a/.github/workflows/test-unit-integrations.yml b/.github/workflows/test-unit-integrations.yml index b28ba3456ce..c95ed4e7c24 100644 --- a/.github/workflows/test-unit-integrations.yml +++ b/.github/workflows/test-unit-integrations.yml @@ -7,10 +7,6 @@ on: - litellm_internal_staging - litellm_oss_staging - "litellm_**" - paths-ignore: - - "ui/**" - - "**.md" - - "**.mdx" permissions: contents: read diff --git a/.github/workflows/test-unit-llm-providers.yml b/.github/workflows/test-unit-llm-providers.yml index fecdcbd3b95..df78564ab0c 100644 --- a/.github/workflows/test-unit-llm-providers.yml +++ b/.github/workflows/test-unit-llm-providers.yml @@ -7,10 +7,6 @@ on: - litellm_internal_staging - litellm_oss_staging - "litellm_**" - paths-ignore: - - "ui/**" - - "**.md" - - "**.mdx" permissions: contents: read diff --git a/.github/workflows/test-unit-misc.yml b/.github/workflows/test-unit-misc.yml index dbc3bfc8191..7c3b195f0ad 100644 --- a/.github/workflows/test-unit-misc.yml +++ b/.github/workflows/test-unit-misc.yml @@ -7,10 +7,6 @@ on: - litellm_internal_staging - litellm_oss_staging - "litellm_**" - paths-ignore: - - "ui/**" - - "**.md" - - "**.mdx" permissions: contents: read diff --git a/.github/workflows/test-unit-proxy-auth.yml b/.github/workflows/test-unit-proxy-auth.yml index ad534cc0098..97dfaed6e81 100644 --- a/.github/workflows/test-unit-proxy-auth.yml +++ b/.github/workflows/test-unit-proxy-auth.yml @@ -7,10 +7,6 @@ on: - litellm_internal_staging - litellm_oss_staging - "litellm_**" - paths-ignore: - - "ui/**" - - "**.md" - - "**.mdx" permissions: contents: read diff --git a/.github/workflows/test-unit-proxy-db.yml b/.github/workflows/test-unit-proxy-db.yml index 35a1a9c78a0..2ac9a3b7c1c 100644 --- a/.github/workflows/test-unit-proxy-db.yml +++ b/.github/workflows/test-unit-proxy-db.yml @@ -5,10 +5,6 @@ on: branches: - main - litellm_internal_staging - paths-ignore: - - "ui/**" - - "**.md" - - "**.mdx" permissions: contents: read diff --git a/.github/workflows/test-unit-proxy-endpoints.yml b/.github/workflows/test-unit-proxy-endpoints.yml index 7eb3d7719c0..cbb36eebdb9 100644 --- a/.github/workflows/test-unit-proxy-endpoints.yml +++ b/.github/workflows/test-unit-proxy-endpoints.yml @@ -7,10 +7,6 @@ on: - litellm_internal_staging - litellm_oss_staging - "litellm_**" - paths-ignore: - - "ui/**" - - "**.md" - - "**.mdx" workflow_dispatch: permissions: diff --git a/.github/workflows/test-unit-proxy-infra.yml b/.github/workflows/test-unit-proxy-infra.yml index cb944de5cf9..884d62289b9 100644 --- a/.github/workflows/test-unit-proxy-infra.yml +++ b/.github/workflows/test-unit-proxy-infra.yml @@ -7,10 +7,6 @@ on: - litellm_internal_staging - litellm_oss_staging - "litellm_**" - paths-ignore: - - "ui/**" - - "**.md" - - "**.mdx" permissions: contents: read diff --git a/.github/workflows/test-unit-proxy-legacy.yml b/.github/workflows/test-unit-proxy-legacy.yml index 9798a4e2277..0068e80e584 100644 --- a/.github/workflows/test-unit-proxy-legacy.yml +++ b/.github/workflows/test-unit-proxy-legacy.yml @@ -7,10 +7,6 @@ on: - litellm_internal_staging - litellm_oss_staging - "litellm_**" - paths-ignore: - - "ui/**" - - "**.md" - - "**.mdx" permissions: contents: read @@ -53,6 +49,10 @@ jobs: with: persist-credentials: false + - name: Detect backend-relevant changes + id: changes + uses: ./.github/actions/detect-backend-changes + - name: Set up Python uses: actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065 # v5.6.0 with: @@ -74,16 +74,19 @@ jobs: ${{ runner.os }}-uv- - name: Install dependencies + if: steps.changes.outputs.decision != 'skip' run: | .github/scripts/uv_sync_with_retries.sh --frozen --group ci --group proxy-dev --extra google --extra proxy --extra semantic-router - name: Generate Prisma client + if: steps.changes.outputs.decision != 'skip' env: PRISMA_BINARY_CACHE_DIR: ${{ runner.temp }}/prisma-cache run: | uv run --no-sync prisma generate --schema litellm/proxy/schema.prisma - name: Run tests - ${{ matrix.test-group.name }} + if: steps.changes.outputs.decision != 'skip' env: TEST_PATH: ${{ matrix.test-group.path }} run: | diff --git a/.github/workflows/test-unit-responses-caching-types.yml b/.github/workflows/test-unit-responses-caching-types.yml index 7331544de24..2f177587997 100644 --- a/.github/workflows/test-unit-responses-caching-types.yml +++ b/.github/workflows/test-unit-responses-caching-types.yml @@ -7,10 +7,6 @@ on: - litellm_internal_staging - litellm_oss_staging - "litellm_**" - paths-ignore: - - "ui/**" - - "**.md" - - "**.mdx" permissions: contents: read diff --git a/enterprise/README.md b/enterprise/README.md index f5eb5078e81..c708dad5a06 100644 --- a/enterprise/README.md +++ b/enterprise/README.md @@ -6,4 +6,4 @@ Code in this folder is licensed under a commercial license. Please review the [L 👉 **Using in an Enterprise / Need specific features ?** Meet with us [here](https://enterprise.litellm.ai/demo?month=2024-02) -See all Enterprise Features here 👉 [Docs](https://docs.litellm.ai/docs/proxy/enterprise) +See all Enterprise Features here 👉 [Docs](https://docs.litellm.ai/docs/enterprise) diff --git a/enterprise/litellm_enterprise/enterprise_callbacks/send_emails/base_email.py b/enterprise/litellm_enterprise/enterprise_callbacks/send_emails/base_email.py index be80a12c80a..e7898cac565 100644 --- a/enterprise/litellm_enterprise/enterprise_callbacks/send_emails/base_email.py +++ b/enterprise/litellm_enterprise/enterprise_callbacks/send_emails/base_email.py @@ -919,9 +919,9 @@ class BaseEmailLogger(CustomLogger): """ Construct invitation link for the user - # http://localhost:4000/ui?invitation_id=7a096b3a-37c6-440f-9dd1-ba22e8043f6b + # http://localhost:4000/ui/onboarding?invitation_id=7a096b3a-37c6-440f-9dd1-ba22e8043f6b """ - return f"{base_url}/ui?invitation_id={invitation_id}" + return f"{base_url}/ui/onboarding?invitation_id={invitation_id}" async def send_email( self, diff --git a/enterprise/pyproject.toml b/enterprise/pyproject.toml index b3864ce7878..85ccbef752f 100644 --- a/enterprise/pyproject.toml +++ b/enterprise/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "litellm-enterprise" -version = "0.1.48" +version = "0.1.49" description = "Package for LiteLLM Enterprise features" readme = "README.md" requires-python = ">=3.9" @@ -26,7 +26,7 @@ required-version = ">=0.10.9" module-root = "" [tool.commitizen] -version = "0.1.48" +version = "0.1.49" version_files = [ "pyproject.toml:^version", "../pyproject.toml:litellm-enterprise==", diff --git a/litellm/anthropic_beta_headers_config.json b/litellm/anthropic_beta_headers_config.json index 11fdb26e42d..3f6817f6e35 100644 --- a/litellm/anthropic_beta_headers_config.json +++ b/litellm/anthropic_beta_headers_config.json @@ -102,7 +102,7 @@ "computer-use-2025-01-24": "computer-use-2025-01-24", "computer-use-2025-11-24": "computer-use-2025-11-24", "context-1m-2025-08-07": "context-1m-2025-08-07", - "context-management-2025-06-27": null, + "context-management-2025-06-27": "context-management-2025-06-27", "effort-2025-11-24": "effort-2025-11-24", "fast-mode-2026-02-01": null, "files-api-2025-04-14": null, diff --git a/litellm/constants.py b/litellm/constants.py index 7423d9b2211..715d57e594d 100644 --- a/litellm/constants.py +++ b/litellm/constants.py @@ -715,6 +715,7 @@ openai_compatible_endpoints: List = [ "https://api.clarifai.com/v2/ext/openai/v1", "https://api.libertai.io/v1", "https://pinstripes.io/v1", + "https://api.meta.ai/v1", ] @@ -781,6 +782,7 @@ openai_compatible_providers: List = [ "ragflow", "pinstripes", # Pinstripes - JSON-configured provider "darkbloom", + "meta", # Meta Model API (Muse Spark) - JSON-configured provider ] openai_text_completion_compatible_providers: List = [ # providers that support `/v1/completions` "together_ai", diff --git a/litellm/exceptions.py b/litellm/exceptions.py index adf7b3ef05a..aca3fb551cc 100644 --- a/litellm/exceptions.py +++ b/litellm/exceptions.py @@ -1180,20 +1180,6 @@ class ModifyResponseException(Exception): super().__init__(message) -class GuardrailInterventionNormalStringError( - Exception -): # custom exception to raise when a guardrail intervenes, but we want to return a normal string to the user - def __init__(self, message: str): - self.message = message - super().__init__(self.message) - - def __str__(self): - return self.message - - def __repr__(self): - return self.__str__() - - class SensitiveDataRouteException(Exception): """ Exception raised when a guardrail detects sensitive data and wants to reroute the request. diff --git a/litellm/integrations/custom_guardrail.py b/litellm/integrations/custom_guardrail.py index 59d37639098..ae82e7992ed 100644 --- a/litellm/integrations/custom_guardrail.py +++ b/litellm/integrations/custom_guardrail.py @@ -757,6 +757,12 @@ class CustomGuardrail(CustomLogger): # raw provider JSON so redaction is not duplicated upstream). clean_guardrail_response = redact_nested_match_and_regex_keys(clean_guardrail_response) + from litellm.litellm_core_utils.sensitive_data_masker import ( + mask_credentials_in_payload, + ) + + clean_guardrail_response = mask_credentials_in_payload(clean_guardrail_response) + slg = StandardLoggingGuardrailInformation( guardrail_name=self.guardrail_name, guardrail_provider=guardrail_provider, diff --git a/litellm/integrations/otel/__init__.py b/litellm/integrations/otel/__init__.py index 7f78f7156b4..5e167e006ff 100644 --- a/litellm/integrations/otel/__init__.py +++ b/litellm/integrations/otel/__init__.py @@ -52,6 +52,7 @@ from litellm.integrations.otel.model.semconv import ( GenAIProvider, JsonRpc, LiteLLM, + LiteLLMError, MCPMethod, Metric, Network, @@ -87,6 +88,7 @@ __all__ = [ "HTTP", "JsonRpc", "LiteLLM", + "LiteLLMError", "MCP", "MCPMethod", "Metric", diff --git a/litellm/integrations/otel/emitter.py b/litellm/integrations/otel/emitter.py index 8441cbae834..46aa166a8bb 100644 --- a/litellm/integrations/otel/emitter.py +++ b/litellm/integrations/otel/emitter.py @@ -16,9 +16,10 @@ from litellm.integrations.otel.model.payloads import ( MCPListToolsSpanData, MCPToolCallSpanData, ServiceSpanData, + SpanError, ) from litellm.integrations.otel.plumbing.providers import to_otel_span_kind -from litellm.integrations.otel.model.semconv import Error, ExceptionEvent +from litellm.integrations.otel.model.semconv import Error, ExceptionEvent, LiteLLMError from litellm.integrations.otel.model.spans import ( SPAN_REGISTRY, SpanRole, @@ -49,6 +50,27 @@ _NAME_BUILDERS: dict[SpanRole, Callable[..., str]] = { _DEDUP_CACHE_MAX = 10_000 +def _stamp_otel_error_attributes(span: Span, error_type: str, resolved_message: str) -> None: + """Stamp the OTel-semconv error attributes (``error.type`` + ``error.message``). + ``error_type`` and ``resolved_message`` are ``finish_span``'s already-computed + fallback chains, so the pair on the status, event, and attributes stays in + lockstep.""" + span.set_attribute(Error.TYPE, error_type) + span.set_attribute(Error.MESSAGE, resolved_message) + + +def _stamp_litellm_error_attributes(span: Span, error: SpanError) -> None: + """Stamp litellm-specific error detail attributes. Emitted only when the + corresponding field is populated so guardrail-shape errors carrying only a + message aren't polluted with empty detail keys.""" + if error.code: + span.set_attribute(LiteLLMError.CODE, error.code) + if error.stack_trace: + span.set_attribute(LiteLLMError.STACK_TRACE, error.stack_trace) + if error.llm_provider: + span.set_attribute(LiteLLMError.LLM_PROVIDER, error.llm_provider) + + class SpanEmitter: def __init__( self, @@ -190,12 +212,13 @@ class SpanEmitter: if error and (error.error_type or error.message): error_type = error.error_type or "error" message = error.message or error.error_type or "error" - span.set_attribute(Error.TYPE, error_type) + _stamp_otel_error_attributes(span, error_type, message) + _stamp_litellm_error_attributes(span, error) span.set_status(Status(StatusCode.ERROR, message)) - # Carry the full message on the standard ``exception`` event so backends - # map it as full text under ``exception.message``. Setting it as a bare - # string attribute instead lets backends like Elasticsearch dynamic-map - # it to a ``keyword`` capped at 1024 chars, truncating the message. + # Also emit the semconv ``exception`` event so backends that + # dynamic-map unknown string span attrs to ``keyword`` (e.g. + # Elasticsearch with a 1024-char ``ignore_above``) still see the + # full untruncated message on the recognized event field. span.add_event( ExceptionEvent.NAME, {ExceptionEvent.TYPE: error_type, ExceptionEvent.MESSAGE: message}, diff --git a/litellm/integrations/otel/model/payloads.py b/litellm/integrations/otel/model/payloads.py index fcd710492f0..4a8f01858b5 100644 --- a/litellm/integrations/otel/model/payloads.py +++ b/litellm/integrations/otel/model/payloads.py @@ -141,6 +141,9 @@ class LLMCost: class SpanError: error_type: str | None = None message: str | None = None + code: str | None = None + stack_trace: str | None = None + llm_provider: str | None = None @dataclass(frozen=True) @@ -571,6 +574,9 @@ def _parse_error(payload: "StandardLoggingPayload") -> SpanError | None: return SpanError( error_type=as_str(info.get("error_class")) or as_str(info.get("error_code")), message=as_str(info.get("error_message")) or as_str(payload.get("error_str")), + code=as_str(info.get("error_code")), + stack_trace=as_str(info.get("traceback")), + llm_provider=as_str(info.get("llm_provider")), ) diff --git a/litellm/integrations/otel/model/semconv.py b/litellm/integrations/otel/model/semconv.py index 4e725ae0a29..aab80c7e5c4 100644 --- a/litellm/integrations/otel/model/semconv.py +++ b/litellm/integrations/otel/model/semconv.py @@ -144,7 +144,24 @@ class Client: class Error: + """OTel-defined error attribute keys, from the semconv ``error.*`` registry. + ``MESSAGE`` is marked *Deprecated* upstream in favor of domain-specific + error message keys plus ``exception.message`` on the exception event, but + litellm still stamps it.""" + TYPE: Final = "error.type" + MESSAGE: Final = "error.message" + + +class LiteLLMError: + """Detail keys for the mapped provider exception of a failed LLM call. + OTel semconv does not define these, so they live under the ``litellm.*`` + vendor namespace rather than squatting on the semconv-owned ``error.*`` + namespace.""" + + CODE: Final = "litellm.provider.error.code" + STACK_TRACE: Final = "litellm.provider.error.stack_trace" + LLM_PROVIDER: Final = "litellm.provider.error.llm_provider" class ExceptionEvent: diff --git a/litellm/litellm_core_utils/get_llm_provider_logic.py b/litellm/litellm_core_utils/get_llm_provider_logic.py index 61a73201c43..fa8ffe1ba1d 100644 --- a/litellm/litellm_core_utils/get_llm_provider_logic.py +++ b/litellm/litellm_core_utils/get_llm_provider_logic.py @@ -346,6 +346,9 @@ def get_llm_provider( elif endpoint == "https://pinstripes.io/v1": custom_llm_provider = "pinstripes" dynamic_api_key = get_secret_str("PINSTRIPES_API_KEY") + elif endpoint == "https://api.meta.ai/v1": + custom_llm_provider = "meta" + dynamic_api_key = get_secret_str("META_API_KEY") if api_base is not None and not isinstance(api_base, str): raise Exception("api base needs to be a string. api_base={}".format(api_base)) diff --git a/litellm/litellm_core_utils/health_check_helpers.py b/litellm/litellm_core_utils/health_check_helpers.py index 405366382a1..9fc036e2a99 100644 --- a/litellm/litellm_core_utils/health_check_helpers.py +++ b/litellm/litellm_core_utils/health_check_helpers.py @@ -95,6 +95,17 @@ class HealthCheckHelpers: """ import litellm + logging_obj = filtered_model_params.get("litellm_logging_obj") + if logging_obj is not None: + api_base = filtered_model_params.get("api_base") + logging_obj.update_from_kwargs( + kwargs=filtered_model_params, + model=filtered_model_params.get("model"), + user=None, + optional_params={}, + litellm_params={"api_base": api_base} if api_base else None, + ) + if custom_llm_provider in LIST_BATCHES_SUPPORTED_PROVIDERS: return await litellm.alist_batches(**filtered_model_params) else: @@ -188,6 +199,7 @@ class HealthCheckHelpers: api_base=model_params.get("api_base", None), api_key=model_params.get("api_key", None), api_version=model_params.get("api_version", None), + model_params=model_params, ), "batch": lambda: HealthCheckHelpers._batch_health_check( custom_llm_provider=custom_llm_provider, diff --git a/litellm/litellm_core_utils/prompt_templates/factory.py b/litellm/litellm_core_utils/prompt_templates/factory.py index c1635158d3b..8bb0e12905e 100644 --- a/litellm/litellm_core_utils/prompt_templates/factory.py +++ b/litellm/litellm_core_utils/prompt_templates/factory.py @@ -3626,6 +3626,7 @@ class BedrockImageProcessor: def _convert_to_bedrock_tool_call_invoke( tool_calls: list, + model: Optional[str] = None, ) -> List[BedrockContentBlock]: """ OpenAI tool invokes: @@ -3701,7 +3702,13 @@ def _convert_to_bedrock_tool_call_invoke( # cache_control applies to the whole original # tool call; attach after the last split block. if tool.get("cache_control", None) is not None: - _parts_list.append(BedrockContentBlock(cachePoint=CachePointBlock(type="default"))) + _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block( + {"cache_control": tool["cache_control"]}, + block_type="content_block", + model=model, + ) + if _cache_point_block is not None: + _parts_list.append(_cache_point_block) continue # Fallback: no objects extracted — use empty dict. arguments_dict = {} @@ -3712,8 +3719,13 @@ def _convert_to_bedrock_tool_call_invoke( # Check for cache_control and add a separate cachePoint block if tool.get("cache_control", None) is not None: - cache_point_block = BedrockContentBlock(cachePoint=CachePointBlock(type="default")) - _parts_list.append(cache_point_block) + cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block( + {"cache_control": tool["cache_control"]}, + block_type="content_block", + model=model, + ) + if cache_point_block is not None: + _parts_list.append(cache_point_block) return _parts_list except Exception as e: raise Exception( @@ -4377,6 +4389,7 @@ class BedrockConverseMessagesProcessor: _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block( message_block=cast(OpenAIMessageContentListBlock, element), block_type="content_block", + model=model, ) if _cache_point_block is not None: _parts.append(_cache_point_block) @@ -4384,7 +4397,7 @@ class BedrockConverseMessagesProcessor: elif message_block["content"] and isinstance(message_block["content"], str): _part = BedrockContentBlock(text=messages[msg_i]["content"]) _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block( - message_block, block_type="content_block" + message_block, block_type="content_block", model=model ) user_content.append(_part) if _cache_point_block is not None: @@ -4416,22 +4429,27 @@ class BedrockConverseMessagesProcessor: tool_content.append(tool_call_result) # Check if we need to add a separate cachePoint block - has_cache_control = False + tool_msg_cache_control = None # Check for message-level cache_control if current_message.get("cache_control", None) is not None: - has_cache_control = True + tool_msg_cache_control = current_message["cache_control"] # Check for content-level cache_control in list content elif isinstance(current_message.get("content"), list): for content_element in current_message["content"]: if isinstance(content_element, dict) and content_element.get("cache_control", None) is not None: - has_cache_control = True + tool_msg_cache_control = content_element["cache_control"] break # Add a separate cachePoint block if cache_control is present - if has_cache_control: - cache_point_block = BedrockContentBlock(cachePoint=CachePointBlock(type="default")) - tool_content.append(cache_point_block) + if tool_msg_cache_control is not None: + cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block( + {"cache_control": tool_msg_cache_control}, + block_type="content_block", + model=model, + ) + if cache_point_block is not None: + tool_content.append(cache_point_block) msg_i += 1 # Deduplicate toolResult blocks with the same toolUseId @@ -4509,6 +4527,7 @@ class BedrockConverseMessagesProcessor: _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block( message_block=cast(OpenAIMessageContentListBlock, element), block_type="content_block", + model=model, ) if _cache_point_block is not None: assistants_parts.append(_cache_point_block) @@ -4520,14 +4539,14 @@ class BedrockConverseMessagesProcessor: # If content is empty/whitespace, skip it (don't add a placeholder) # Add cache point block for assistant string content _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block( - assistant_message_block, block_type="content_block" + assistant_message_block, block_type="content_block", model=model ) if _cache_point_block is not None: assistant_content.append(_cache_point_block) _tool_calls = assistant_message_block.get("tool_calls", []) if _tool_calls: - assistant_content.extend(_convert_to_bedrock_tool_call_invoke(_tool_calls)) + assistant_content.extend(_convert_to_bedrock_tool_call_invoke(_tool_calls, model=model)) msg_i += 1 @@ -4745,6 +4764,7 @@ def _bedrock_converse_messages_pt( _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block( message_block=cast(OpenAIMessageContentListBlock, element), block_type="content_block", + model=model, ) if _cache_point_block is not None: _parts.append(_cache_point_block) @@ -4752,7 +4772,7 @@ def _bedrock_converse_messages_pt( elif message_block["content"] and isinstance(message_block["content"], str): _part = BedrockContentBlock(text=messages[msg_i]["content"]) _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block( - message_block, block_type="content_block" + message_block, block_type="content_block", model=model ) user_content.append(_part) if _cache_point_block is not None: @@ -4786,22 +4806,27 @@ def _bedrock_converse_messages_pt( tool_content.append(tool_call_result) # Check if we need to add a separate cachePoint block - has_cache_control = False + tool_msg_cache_control = None # Check for message-level cache_control if current_message.get("cache_control", None) is not None: - has_cache_control = True + tool_msg_cache_control = current_message["cache_control"] # Check for content-level cache_control in list content elif isinstance(current_message.get("content"), list): for content_element in current_message["content"]: if isinstance(content_element, dict) and content_element.get("cache_control", None) is not None: - has_cache_control = True + tool_msg_cache_control = content_element["cache_control"] break # Add a separate cachePoint block if cache_control is present - if has_cache_control: - cache_point_block = BedrockContentBlock(cachePoint=CachePointBlock(type="default")) - tool_content.append(cache_point_block) + if tool_msg_cache_control is not None: + cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block( + {"cache_control": tool_msg_cache_control}, + block_type="content_block", + model=model, + ) + if cache_point_block is not None: + tool_content.append(cache_point_block) msg_i += 1 # Deduplicate toolResult blocks with the same toolUseId @@ -4882,6 +4907,7 @@ def _bedrock_converse_messages_pt( _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block( message_block=cast(OpenAIMessageContentListBlock, element), block_type="content_block", + model=model, ) if _cache_point_block is not None: assistants_parts.append(_cache_point_block) @@ -4892,13 +4918,13 @@ def _bedrock_converse_messages_pt( assistant_content.append(BedrockContentBlock(text=_assistant_content)) # Add cache point block for assistant string content _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block( - assistant_message_block, block_type="content_block" + assistant_message_block, block_type="content_block", model=model ) if _cache_point_block is not None: assistant_content.append(_cache_point_block) _tool_calls = assistant_message_block.get("tool_calls", []) if _tool_calls: - assistant_content.extend(_convert_to_bedrock_tool_call_invoke(_tool_calls)) + assistant_content.extend(_convert_to_bedrock_tool_call_invoke(_tool_calls, model=model)) msg_i += 1 diff --git a/litellm/litellm_core_utils/sensitive_data_masker.py b/litellm/litellm_core_utils/sensitive_data_masker.py index 1f3a6961f39..7861e13bae5 100644 --- a/litellm/litellm_core_utils/sensitive_data_masker.py +++ b/litellm/litellm_core_utils/sensitive_data_masker.py @@ -1,6 +1,8 @@ from collections.abc import Mapping from typing import Any, Dict, List, Optional, Set +from pydantic import BaseModel + from litellm.constants import DEFAULT_MAX_RECURSE_DEPTH_SENSITIVE_DATA_MASKER @@ -153,6 +155,39 @@ def mask_sensitive_structure(data: object) -> object: return _error_masker.mask(data) +def mask_credentials_in_payload(data: object) -> object: + """Return a copy of ``data`` where string values under sensitive-named keys + are masked but every other value (``None``, ``int``, ``float``, ``bool``, + ``bytes``, ``datetime``, tuples, sets, typed objects) is preserved by + identity, and dicts/lists are rebuilt structurally. + + Use this for logging payloads that carry response data through to + SpendLogs / OTel / Langfuse, where :meth:`SensitiveDataMasker.mask`'s + config-dump semantics (``None`` -> ``"None"``, tuples stringified, + objects flattened via ``__dict__``) would silently distort the record. + + Sensitive-key detection is delegated to the shared + :class:`SensitiveDataMasker` so pattern updates stay in one place. + """ + return _walk_payload(data, key_is_sensitive=False, depth=0) + + +def _walk_payload(node: object, key_is_sensitive: bool, depth: int) -> object: + if depth >= DEFAULT_MAX_RECURSE_DEPTH_SENSITIVE_DATA_MASKER: + return node + if isinstance(node, Mapping): + return {k: _walk_payload(v, _default_masker.is_sensitive_key(k), depth + 1) for k, v in node.items()} + if isinstance(node, list): + return [_walk_payload(item, key_is_sensitive, depth + 1) for item in node] + if isinstance(node, tuple): + return tuple(_walk_payload(item, key_is_sensitive, depth + 1) for item in node) + if isinstance(node, BaseModel): + return _walk_payload(node.model_dump(), key_is_sensitive, depth) + if key_is_sensitive and isinstance(node, str) and node: + return _default_masker._mask_value(node) + return node + + def mask_sensitive_keys(data: Dict[str, Any], sensitive_fields: Set[str]) -> Dict[str, Any]: """Return a new dict with values masked for keys listed in ``sensitive_fields``. diff --git a/litellm/llms/azure_ai/ocr/common_utils.py b/litellm/llms/azure_ai/ocr/common_utils.py index d1d5b80b78d..14b77338fd7 100644 --- a/litellm/llms/azure_ai/ocr/common_utils.py +++ b/litellm/llms/azure_ai/ocr/common_utils.py @@ -13,6 +13,17 @@ if TYPE_CHECKING: from litellm.llms.base_llm.ocr.transformation import BaseOCRConfig +def is_azure_document_intelligence_model(model: str) -> bool: + """Whether an azure_ai OCR model routes to Azure Document Intelligence. + + Azure AI exposes two OCR services on the same provider; the sub-route in the + model name (`azure_ai/doc-intelligence/`) selects Document Intelligence + over Mistral OCR. This is the single source of truth for that routing decision. + """ + lowered = model.lower() + return "doc-intelligence" in lowered or "documentintelligence" in lowered + + def get_azure_ai_ocr_config(model: str) -> Optional["BaseOCRConfig"]: """ Determine which Azure AI OCR configuration to use based on the model name. @@ -41,7 +52,7 @@ def get_azure_ai_ocr_config(model: str) -> Optional["BaseOCRConfig"]: from litellm.llms.azure_ai.ocr.transformation import AzureAIOCRConfig # Check for Azure Document Intelligence models - if "doc-intelligence" in model or "documentintelligence" in model: + if is_azure_document_intelligence_model(model): verbose_logger.debug(f"Routing {model} to Azure Document Intelligence OCR config") return AzureDocumentIntelligenceOCRConfig() diff --git a/litellm/llms/bedrock/messages/invoke_transformations/anthropic_claude3_transformation.py b/litellm/llms/bedrock/messages/invoke_transformations/anthropic_claude3_transformation.py index f5309d521a9..daee3369a3c 100644 --- a/litellm/llms/bedrock/messages/invoke_transformations/anthropic_claude3_transformation.py +++ b/litellm/llms/bedrock/messages/invoke_transformations/anthropic_claude3_transformation.py @@ -489,24 +489,43 @@ class AmazonAnthropicClaudeMessagesConfig( if self._supports_tool_search_on_bedrock(model): beta_set.add("tool-search-tool-2025-10-19") + # Bedrock-InvokeModel-supported ``context_management.edits`` types and the + # ``anthropic-beta`` header that each one requires. ``clear_thinking_20251015`` + # is intentionally absent — it is LiteLLM-internal, consumed via + # ``_ensure_thinking_for_clear_thinking_context_management``, and forwarding + # the raw edit trips Bedrock's + # ``"context_management: Extra inputs are not permitted"`` 400. + # + # Bedrock InvokeModel DOES support ``clear_tool_uses_20250919`` under the + # ``context-management-2025-06-27`` beta. AWS docs: + # https://docs.aws.amazon.com/bedrock/latest/userguide/model-parameters-anthropic-claude-messages-tool-use.md + _BEDROCK_INVOKE_SUPPORTED_CONTEXT_MANAGEMENT_EDITS: Dict[str, str] = { + "compact_20260112": ANTHROPIC_BETA_HEADER_VALUES.COMPACT_2026_01_12.value, + "clear_tool_uses_20250919": ANTHROPIC_BETA_HEADER_VALUES.CONTEXT_MANAGEMENT_2025_06_27.value, + } + @staticmethod def _filter_context_management_for_bedrock_invoke( anthropic_messages_request: Dict, beta_set: set, ) -> None: """ - Bedrock InvokeModel accepts ``context_management`` only when it carries - ``compact_20260112`` edits paired with the ``compact-2026-01-12`` - anthropic-beta header. Other edit types (notably ``clear_thinking_20251015``, - which Claude Code sends on every request) are LiteLLM-internal and would - cause Bedrock to 400 with ``"context_management: Extra inputs are not - permitted"``. + Filter ``context_management.edits`` to the subset that Bedrock InvokeModel + accepts and add the matching ``anthropic-beta`` header for each surviving + edit type. - Filter the edits list to the supported subset, add the beta header when - compact edits remain, and drop ``context_management`` entirely when no - supported edits are left so the safety-net allowlist can pass it through. + - ``compact_20260112`` -> ``compact-2026-01-12`` + - ``clear_tool_uses_20250919`` -> ``context-management-2025-06-27`` - Ref: https://github.com/BerriAI/litellm/issues/27532 + Other edit types (notably ``clear_thinking_20251015``, which Claude Code + sends on every request) are LiteLLM-internal: thinking is injected + separately via ``_ensure_thinking_for_clear_thinking_context_management``, + and forwarding the raw edit would trip Bedrock's + ``"context_management: Extra inputs are not permitted"`` 400. + + Refs: + * https://github.com/BerriAI/litellm/issues/27532 + * https://docs.aws.amazon.com/bedrock/latest/userguide/model-parameters-anthropic-claude-messages-tool-use.md """ cm = anthropic_messages_request.get("context_management") if not isinstance(cm, dict): @@ -516,15 +535,17 @@ class AmazonAnthropicClaudeMessagesConfig( anthropic_messages_request.pop("context_management", None) return - compact_edits = [e for e in edits if isinstance(e, dict) and e.get("type") == "compact_20260112"] - if compact_edits: - beta_set.add(ANTHROPIC_BETA_HEADER_VALUES.COMPACT_2026_01_12.value) - anthropic_messages_request["context_management"] = { - **cm, - "edits": compact_edits, - } - else: + supported = AmazonAnthropicClaudeMessagesConfig._BEDROCK_INVOKE_SUPPORTED_CONTEXT_MANAGEMENT_EDITS + retained_edits = [e for e in edits if isinstance(e, dict) and e.get("type") in supported] + if not retained_edits: anthropic_messages_request.pop("context_management", None) + return + + beta_set.update(supported[e["type"]] for e in retained_edits) + anthropic_messages_request["context_management"] = { + **cm, + "edits": retained_edits, + } def _get_bedrock_invoke_anthropic_beta_headers( self, diff --git a/litellm/llms/deepseek/chat/transformation.py b/litellm/llms/deepseek/chat/transformation.py index 7a548136f2a..525de1476e2 100644 --- a/litellm/llms/deepseek/chat/transformation.py +++ b/litellm/llms/deepseek/chat/transformation.py @@ -35,7 +35,9 @@ class DeepSeekChatConfig(OpenAIGPTConfig): Map OpenAI params to DeepSeek params. Handles `thinking` and `reasoning_effort` parameters for DeepSeek reasoner models. - DeepSeek only supports `{"type": "enabled"}` - no budget_tokens like Anthropic. + DeepSeek supports `{"type": "enabled"}` and `{"type": "disabled"}` - no budget_tokens + like Anthropic. `reasoning_effort="none"` is the OpenAI-style way to ask for thinking + off, so it maps to `{"type": "disabled"}`; any other effort keeps thinking on. Reference: https://api-docs.deepseek.com/guides/thinking_mode """ @@ -47,15 +49,13 @@ class DeepSeekChatConfig(OpenAIGPTConfig): thinking_value = optional_params.pop("thinking", None) reasoning_effort = optional_params.pop("reasoning_effort", None) - # Handle thinking parameter - only accept {"type": "enabled"} - if thinking_value is not None: - if isinstance(thinking_value, dict) and thinking_value.get("type") == "enabled": - # DeepSeek only accepts {"type": "enabled"}, ignore budget_tokens - optional_params["thinking"] = {"type": "enabled"} + # Handle thinking parameter - accept both enabled and disabled, ignore budget_tokens + if isinstance(thinking_value, dict) and thinking_value.get("type") in ("enabled", "disabled"): + optional_params["thinking"] = {"type": thinking_value["type"]} - # Handle reasoning_effort - map to thinking enabled - elif reasoning_effort is not None and reasoning_effort != "none": - optional_params["thinking"] = {"type": "enabled"} + # Otherwise fall back to reasoning_effort: "none" disables, anything else enables + elif reasoning_effort is not None: + optional_params["thinking"] = {"type": "disabled" if reasoning_effort == "none" else "enabled"} return optional_params diff --git a/litellm/llms/openai_like/dynamic_config.py b/litellm/llms/openai_like/dynamic_config.py index 3c763ed9b9b..31c913d5d4e 100644 --- a/litellm/llms/openai_like/dynamic_config.py +++ b/litellm/llms/openai_like/dynamic_config.py @@ -91,7 +91,7 @@ def create_config_class(provider: SimpleProviderConfig): def get_supported_openai_params(self, model: str) -> list: """Get supported OpenAI params, excluding tool-related params for models that don't support function calling.""" - from litellm.utils import supports_function_calling + from litellm.utils import supports_function_calling, supports_reasoning supported_params = super().get_supported_openai_params(model=model) @@ -113,6 +113,10 @@ def create_config_class(provider: SimpleProviderConfig): f"function calling — removed tool-related params from supported params." ) + _supports_reasoning = supports_reasoning(model=model, custom_llm_provider=provider.slug) + if _supports_reasoning and "reasoning_effort" not in supported_params: + supported_params.append("reasoning_effort") + return supported_params def map_openai_params( diff --git a/litellm/llms/openai_like/messages/transformation.py b/litellm/llms/openai_like/messages/transformation.py index 0df8c6e830b..4963bcca9ac 100644 --- a/litellm/llms/openai_like/messages/transformation.py +++ b/litellm/llms/openai_like/messages/transformation.py @@ -1,8 +1,11 @@ from typing import Any, Optional +import litellm from litellm.llms.anthropic.experimental_pass_through.messages.transformation import ( AnthropicMessagesConfig, ) +from litellm.llms.openai_like.json_loader import SimpleProviderConfig +from litellm.secret_managers.main import get_secret_str DEFAULT_ANTHROPIC_API_VERSION = "2023-06-01" @@ -67,3 +70,65 @@ class OpenAILikeAnthropicMessagesConfig(AnthropicMessagesConfig): if base.endswith("/v1"): base = base[: -len("/v1")] return f"{base}/v1/messages" + + +class JSONProviderAnthropicMessagesConfig(OpenAILikeAnthropicMessagesConfig): + """ + Provider-level native Anthropic Messages passthrough for JSON-configured + OpenAI-compatible providers whose ``supported_endpoints`` in providers.json + includes ``"/v1/messages"``. Resolves the api key and api base from the + provider's configured env vars, then forwards the Anthropic payload + untranslated like ``OpenAILikeAnthropicMessagesConfig``. + """ + + def __init__(self, provider: SimpleProviderConfig): + super().__init__() + self._provider = provider + + def should_strip_billing_metadata(self) -> bool: + return True + + def _resolve_api_key(self, api_key: Optional[str]) -> Optional[str]: + return api_key or get_secret_str(self._provider.api_key_env) or litellm.api_key + + def _resolve_api_base(self, api_base: Optional[str]) -> str: + env_api_base = get_secret_str(self._provider.api_base_env) if self._provider.api_base_env else None + return api_base or env_api_base or self._provider.base_url + + def validate_anthropic_messages_environment( + self, + headers: dict[str, str], + model: str, + messages: list[Any], + optional_params: dict, + litellm_params: dict, + api_key: Optional[str] = None, + api_base: Optional[str] = None, + ) -> tuple[dict[str, str], Optional[str]]: + return super().validate_anthropic_messages_environment( + headers=headers, + model=model, + messages=messages, + optional_params=optional_params, + litellm_params=litellm_params, + api_key=self._resolve_api_key(api_key), + api_base=api_base, + ) + + def get_complete_url( + self, + api_base: Optional[str], + api_key: Optional[str], + model: str, + optional_params: dict, + litellm_params: dict, + stream: Optional[bool] = None, + ) -> str: + return super().get_complete_url( + api_base=self._resolve_api_base(api_base), + api_key=api_key, + model=model, + optional_params=optional_params, + litellm_params=litellm_params, + stream=stream, + ) diff --git a/litellm/llms/openai_like/providers.json b/litellm/llms/openai_like/providers.json index d87346fea70..164100d4194 100644 --- a/litellm/llms/openai_like/providers.json +++ b/litellm/llms/openai_like/providers.json @@ -168,6 +168,13 @@ }, "supported_endpoints": ["/v1/chat/completions", "/v1/responses"] }, + "meta": { + "base_url": "https://api.meta.ai/v1", + "api_key_env": "META_API_KEY", + "api_base_env": "META_API_BASE", + "base_class": "openai_gpt", + "supported_endpoints": ["/v1/chat/completions", "/v1/responses", "/v1/messages"] + }, "pinstripes": { "base_url": "https://pinstripes.io/v1", "api_key_env": "PINSTRIPES_API_KEY", diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json index 70b6b05e6ec..2a269d693ec 100644 --- a/litellm/model_prices_and_context_window_backup.json +++ b/litellm/model_prices_and_context_window_backup.json @@ -6012,6 +6012,522 @@ "supports_vision": true, "supports_web_search": true }, + "azure/gpt-5.6": { + "cache_read_input_token_cost": 5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1e-06, + "cache_read_input_token_cost_priority": 1e-06, + "cache_read_input_token_cost_above_272k_tokens_priority": 2e-06, + "input_cost_per_token": 5e-06, + "input_cost_per_token_above_272k_tokens": 1e-05, + "input_cost_per_token_priority": 1e-05, + "input_cost_per_token_above_272k_tokens_priority": 2e-05, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3e-05, + "output_cost_per_token_above_272k_tokens": 4.5e-05, + "output_cost_per_token_priority": 6e-05, + "output_cost_per_token_above_272k_tokens_priority": 9e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/gpt-5.6-sol": { + "cache_read_input_token_cost": 5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1e-06, + "cache_read_input_token_cost_priority": 1e-06, + "cache_read_input_token_cost_above_272k_tokens_priority": 2e-06, + "input_cost_per_token": 5e-06, + "input_cost_per_token_above_272k_tokens": 1e-05, + "input_cost_per_token_priority": 1e-05, + "input_cost_per_token_above_272k_tokens_priority": 2e-05, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3e-05, + "output_cost_per_token_above_272k_tokens": 4.5e-05, + "output_cost_per_token_priority": 6e-05, + "output_cost_per_token_above_272k_tokens_priority": 9e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/gpt-5.6-terra": { + "cache_read_input_token_cost": 2.5e-07, + "cache_read_input_token_cost_above_272k_tokens": 5e-07, + "cache_read_input_token_cost_priority": 5e-07, + "cache_read_input_token_cost_above_272k_tokens_priority": 1e-06, + "input_cost_per_token": 2.5e-06, + "input_cost_per_token_above_272k_tokens": 5e-06, + "input_cost_per_token_priority": 5e-06, + "input_cost_per_token_above_272k_tokens_priority": 1e-05, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 1.5e-05, + "output_cost_per_token_above_272k_tokens": 2.25e-05, + "output_cost_per_token_priority": 3e-05, + "output_cost_per_token_above_272k_tokens_priority": 4.5e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/gpt-5.6-luna": { + "cache_read_input_token_cost": 1e-07, + "cache_read_input_token_cost_above_272k_tokens": 2e-07, + "cache_read_input_token_cost_priority": 2e-07, + "cache_read_input_token_cost_above_272k_tokens_priority": 4e-07, + "input_cost_per_token": 1e-06, + "input_cost_per_token_above_272k_tokens": 2e-06, + "input_cost_per_token_priority": 2e-06, + "input_cost_per_token_above_272k_tokens_priority": 4e-06, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 6e-06, + "output_cost_per_token_above_272k_tokens": 9e-06, + "output_cost_per_token_priority": 1.2e-05, + "output_cost_per_token_above_272k_tokens_priority": 1.8e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/us/gpt-5.6": { + "cache_read_input_token_cost": 5.5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1.1e-06, + "cache_read_input_token_cost_priority": 1.375e-06, + "input_cost_per_token": 5.5e-06, + "input_cost_per_token_above_272k_tokens": 1.1e-05, + "input_cost_per_token_priority": 1.375e-05, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3.3e-05, + "output_cost_per_token_above_272k_tokens": 4.95e-05, + "output_cost_per_token_priority": 8.25e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/us/gpt-5.6-sol": { + "cache_read_input_token_cost": 5.5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1.1e-06, + "cache_read_input_token_cost_priority": 1.375e-06, + "input_cost_per_token": 5.5e-06, + "input_cost_per_token_above_272k_tokens": 1.1e-05, + "input_cost_per_token_priority": 1.375e-05, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3.3e-05, + "output_cost_per_token_above_272k_tokens": 4.95e-05, + "output_cost_per_token_priority": 8.25e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/us/gpt-5.6-terra": { + "cache_read_input_token_cost": 2.75e-07, + "cache_read_input_token_cost_above_272k_tokens": 5.5e-07, + "cache_read_input_token_cost_priority": 6.875e-07, + "input_cost_per_token": 2.75e-06, + "input_cost_per_token_above_272k_tokens": 5.5e-06, + "input_cost_per_token_priority": 6.875e-06, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 1.65e-05, + "output_cost_per_token_above_272k_tokens": 2.475e-05, + "output_cost_per_token_priority": 4.125e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/us/gpt-5.6-luna": { + "cache_read_input_token_cost": 1.1e-07, + "cache_read_input_token_cost_above_272k_tokens": 2.2e-07, + "cache_read_input_token_cost_priority": 2.75e-07, + "input_cost_per_token": 1.1e-06, + "input_cost_per_token_above_272k_tokens": 2.2e-06, + "input_cost_per_token_priority": 2.75e-06, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 6.6e-06, + "output_cost_per_token_above_272k_tokens": 9.9e-06, + "output_cost_per_token_priority": 1.65e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/eu/gpt-5.6": { + "cache_read_input_token_cost": 5.5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1.1e-06, + "cache_read_input_token_cost_priority": 1.375e-06, + "input_cost_per_token": 5.5e-06, + "input_cost_per_token_above_272k_tokens": 1.1e-05, + "input_cost_per_token_priority": 1.375e-05, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3.3e-05, + "output_cost_per_token_above_272k_tokens": 4.95e-05, + "output_cost_per_token_priority": 8.25e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/eu/gpt-5.6-sol": { + "cache_read_input_token_cost": 5.5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1.1e-06, + "cache_read_input_token_cost_priority": 1.375e-06, + "input_cost_per_token": 5.5e-06, + "input_cost_per_token_above_272k_tokens": 1.1e-05, + "input_cost_per_token_priority": 1.375e-05, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3.3e-05, + "output_cost_per_token_above_272k_tokens": 4.95e-05, + "output_cost_per_token_priority": 8.25e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/eu/gpt-5.6-terra": { + "cache_read_input_token_cost": 2.75e-07, + "cache_read_input_token_cost_above_272k_tokens": 5.5e-07, + "cache_read_input_token_cost_priority": 6.875e-07, + "input_cost_per_token": 2.75e-06, + "input_cost_per_token_above_272k_tokens": 5.5e-06, + "input_cost_per_token_priority": 6.875e-06, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 1.65e-05, + "output_cost_per_token_above_272k_tokens": 2.475e-05, + "output_cost_per_token_priority": 4.125e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/eu/gpt-5.6-luna": { + "cache_read_input_token_cost": 1.1e-07, + "cache_read_input_token_cost_above_272k_tokens": 2.2e-07, + "cache_read_input_token_cost_priority": 2.75e-07, + "input_cost_per_token": 1.1e-06, + "input_cost_per_token_above_272k_tokens": 2.2e-06, + "input_cost_per_token_priority": 2.75e-06, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 6.6e-06, + "output_cost_per_token_above_272k_tokens": 9.9e-06, + "output_cost_per_token_priority": 1.65e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, "azure/gpt-5.5": { "cache_read_input_token_cost": 5e-07, "cache_read_input_token_cost_above_272k_tokens": 1e-06, @@ -22273,6 +22789,218 @@ "supports_xhigh_reasoning_effort": true, "supports_minimal_reasoning_effort": true }, + "gpt-5.6": { + "cache_creation_input_token_cost": 6.25e-06, + "cache_creation_input_token_cost_above_272k_tokens": 1.25e-05, + "cache_creation_input_token_cost_flex": 3.125e-06, + "cache_creation_input_token_cost_priority": 1.25e-05, + "cache_read_input_token_cost": 5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1e-06, + "cache_read_input_token_cost_flex": 2.5e-07, + "cache_read_input_token_cost_priority": 1e-06, + "input_cost_per_token": 5e-06, + "input_cost_per_token_above_272k_tokens": 1e-05, + "input_cost_per_token_batches": 2.5e-06, + "input_cost_per_token_flex": 2.5e-06, + "input_cost_per_token_priority": 1e-05, + "litellm_provider": "openai", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3e-05, + "output_cost_per_token_above_272k_tokens": 4.5e-05, + "output_cost_per_token_batches": 1.5e-05, + "output_cost_per_token_flex": 1.5e-05, + "output_cost_per_token_priority": 6e-05, + "regional_processing_uplift_multiplier_eu": 1.1, + "regional_processing_uplift_multiplier_us": 1.1, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_minimal_reasoning_effort": false, + "supports_native_streaming": true, + "supports_none_reasoning_effort": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_xhigh_reasoning_effort": true + }, + "gpt-5.6-sol": { + "cache_creation_input_token_cost": 6.25e-06, + "cache_creation_input_token_cost_above_272k_tokens": 1.25e-05, + "cache_creation_input_token_cost_flex": 3.125e-06, + "cache_creation_input_token_cost_priority": 1.25e-05, + "cache_read_input_token_cost": 5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1e-06, + "cache_read_input_token_cost_flex": 2.5e-07, + "cache_read_input_token_cost_priority": 1e-06, + "input_cost_per_token": 5e-06, + "input_cost_per_token_above_272k_tokens": 1e-05, + "input_cost_per_token_batches": 2.5e-06, + "input_cost_per_token_flex": 2.5e-06, + "input_cost_per_token_priority": 1e-05, + "litellm_provider": "openai", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3e-05, + "output_cost_per_token_above_272k_tokens": 4.5e-05, + "output_cost_per_token_batches": 1.5e-05, + "output_cost_per_token_flex": 1.5e-05, + "output_cost_per_token_priority": 6e-05, + "regional_processing_uplift_multiplier_eu": 1.1, + "regional_processing_uplift_multiplier_us": 1.1, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_minimal_reasoning_effort": false, + "supports_native_streaming": true, + "supports_none_reasoning_effort": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_xhigh_reasoning_effort": true + }, + "gpt-5.6-terra": { + "cache_creation_input_token_cost": 3.125e-06, + "cache_creation_input_token_cost_above_272k_tokens": 6.25e-06, + "cache_creation_input_token_cost_flex": 1.5625e-06, + "cache_creation_input_token_cost_priority": 6.25e-06, + "cache_read_input_token_cost": 2.5e-07, + "cache_read_input_token_cost_above_272k_tokens": 5e-07, + "cache_read_input_token_cost_flex": 1.25e-07, + "cache_read_input_token_cost_priority": 5e-07, + "input_cost_per_token": 2.5e-06, + "input_cost_per_token_above_272k_tokens": 5e-06, + "input_cost_per_token_batches": 1.25e-06, + "input_cost_per_token_flex": 1.25e-06, + "input_cost_per_token_priority": 5e-06, + "litellm_provider": "openai", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 1.5e-05, + "output_cost_per_token_above_272k_tokens": 2.25e-05, + "output_cost_per_token_batches": 7.5e-06, + "output_cost_per_token_flex": 7.5e-06, + "output_cost_per_token_priority": 3e-05, + "regional_processing_uplift_multiplier_eu": 1.1, + "regional_processing_uplift_multiplier_us": 1.1, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_minimal_reasoning_effort": false, + "supports_native_streaming": true, + "supports_none_reasoning_effort": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_xhigh_reasoning_effort": true + }, + "gpt-5.6-luna": { + "cache_creation_input_token_cost": 1.25e-06, + "cache_creation_input_token_cost_above_272k_tokens": 2.5e-06, + "cache_creation_input_token_cost_flex": 6.25e-07, + "cache_creation_input_token_cost_priority": 2.5e-06, + "cache_read_input_token_cost": 1e-07, + "cache_read_input_token_cost_above_272k_tokens": 2e-07, + "cache_read_input_token_cost_flex": 5e-08, + "cache_read_input_token_cost_priority": 2e-07, + "input_cost_per_token": 1e-06, + "input_cost_per_token_above_272k_tokens": 2e-06, + "input_cost_per_token_batches": 5e-07, + "input_cost_per_token_flex": 5e-07, + "input_cost_per_token_priority": 2e-06, + "litellm_provider": "openai", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 6e-06, + "output_cost_per_token_above_272k_tokens": 9e-06, + "output_cost_per_token_batches": 3e-06, + "output_cost_per_token_flex": 3e-06, + "output_cost_per_token_priority": 1.2e-05, + "regional_processing_uplift_multiplier_eu": 1.1, + "regional_processing_uplift_multiplier_us": 1.1, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_minimal_reasoning_effort": false, + "supports_native_streaming": true, + "supports_none_reasoning_effort": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_xhigh_reasoning_effort": true + }, "gpt-5.5": { "cache_read_input_token_cost": 5e-07, "cache_read_input_token_cost_above_272k_tokens": 1e-06, @@ -25289,6 +26017,42 @@ "supports_function_calling": true, "supports_tool_choice": false }, + "meta/muse-spark-1.1": { + "cache_read_input_token_cost": 1.5e-07, + "input_cost_per_token": 1.25e-06, + "litellm_provider": "meta", + "max_input_tokens": 1048576, + "max_output_tokens": 131072, + "max_tokens": 131072, + "mode": "chat", + "output_cost_per_token": 4.25e-06, + "source": "https://dev.meta.ai/docs/getting-started/pricing-rate-limits", + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/responses", + "/v1/messages" + ], + "supported_modalities": [ + "text", + "image", + "video" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_minimal_reasoning_effort": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_xhigh_reasoning_effort": true + }, "meta_llama/Llama-3.3-70B-Instruct": { "litellm_provider": "meta_llama", "max_input_tokens": 128000, @@ -38101,6 +38865,48 @@ "supports_vision": true, "supports_web_search": true }, + "xai/grok-4.5": { + "cache_read_input_token_cost": 5e-07, + "cache_read_input_token_cost_above_200k_tokens": 1e-06, + "input_cost_per_token": 2e-06, + "input_cost_per_token_above_200k_tokens": 4e-06, + "litellm_provider": "xai", + "max_input_tokens": 500000, + "max_output_tokens": 500000, + "max_tokens": 500000, + "mode": "chat", + "output_cost_per_token": 6e-06, + "output_cost_per_token_above_200k_tokens": 1.2e-05, + "source": "https://docs.x.ai/docs/models", + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true + }, + "xai/grok-4.5-latest": { + "cache_read_input_token_cost": 5e-07, + "cache_read_input_token_cost_above_200k_tokens": 1e-06, + "input_cost_per_token": 2e-06, + "input_cost_per_token_above_200k_tokens": 4e-06, + "litellm_provider": "xai", + "max_input_tokens": 500000, + "max_output_tokens": 500000, + "max_tokens": 500000, + "mode": "chat", + "output_cost_per_token": 6e-06, + "output_cost_per_token_above_200k_tokens": 1.2e-05, + "source": "https://docs.x.ai/docs/models", + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true + }, "xai/grok-beta": { "input_cost_per_token": 5e-06, "litellm_provider": "xai", diff --git a/litellm/ocr/main.py b/litellm/ocr/main.py index 5716155361d..38f3f804e10 100644 --- a/litellm/ocr/main.py +++ b/litellm/ocr/main.py @@ -17,6 +17,9 @@ import litellm from litellm._logging import verbose_logger from litellm.constants import request_timeout from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj +from litellm.llms.azure_ai.ocr.common_utils import ( + is_azure_document_intelligence_model, +) from litellm.llms.base_llm.ocr.transformation import BaseOCRConfig, OCRResponse from litellm.llms.custom_httpx.llm_http_handler import BaseLLMHTTPHandler from litellm.rust_bridge import ocr as rust_ocr_bridge @@ -83,6 +86,8 @@ def _prepare_ocr_request( if doc_type not in ["document_url", "image_url"]: raise ValueError(f"Invalid document type: {doc_type}. Must be 'document_url', 'image_url', or 'file'") + caller_supplied_api_base = api_base is not None + ( model, custom_llm_provider, @@ -95,9 +100,14 @@ def _prepare_ocr_request( api_key=api_key, ) + suppress_dynamic_api_base = ( + not caller_supplied_api_base + and custom_llm_provider == "azure_ai" + and is_azure_document_intelligence_model(model) + ) if dynamic_api_key: api_key = dynamic_api_key - if dynamic_api_base: + if dynamic_api_base and not suppress_dynamic_api_base: api_base = dynamic_api_base ocr_provider_config = ProviderConfigManager.get_provider_ocr_config( @@ -191,8 +201,7 @@ def _rust_bridge_api_base( if prepared_request.api_base is not None: return prepared_request.api_base if prepared_request.custom_llm_provider == "azure_ai": - model = prepared_request.model.lower() - if "doc-intelligence" in model or "documentintelligence" in model: + if is_azure_document_intelligence_model(prepared_request.model): return resolve_secret("AZURE_DOCUMENT_INTELLIGENCE_ENDPOINT") return resolve_secret("AZURE_AI_API_BASE") return None diff --git a/litellm/proxy/_experimental/mcp_server/auth/user_api_key_auth_mcp.py b/litellm/proxy/_experimental/mcp_server/auth/user_api_key_auth_mcp.py index d2986a3cd82..e7ffef0e4e3 100644 --- a/litellm/proxy/_experimental/mcp_server/auth/user_api_key_auth_mcp.py +++ b/litellm/proxy/_experimental/mcp_server/auth/user_api_key_auth_mcp.py @@ -220,6 +220,12 @@ class MCPRequestHandler: # when EVERY target is auth_type=oauth2 with delegate_auth_to_upstream # set; fails closed otherwise. validated_user_api_key_auth = UserAPIKeyAuth() + elif MCPRequestHandler._target_servers_are_true_passthrough( + path=request_route, + mcp_servers=mcp_servers, + client_ip=IPAddressUtils.get_mcp_client_ip(request), + ): + validated_user_api_key_auth = UserAPIKeyAuth() elif oauth2_headers: # Authorization on a non-delegated server: the bearer must be a real # LiteLLM credential, so a failed validation is a genuine 401/403 and @@ -399,6 +405,33 @@ class MCPRequestHandler: return False return True + @staticmethod + def _target_servers_are_true_passthrough( + path: str, mcp_servers: Optional[list[str]], client_ip: Optional[str] + ) -> bool: + """ + True only when EVERY MCP server the request targets is ``auth_type == true_passthrough``. + Fails closed when any target does not opt in or cannot be resolved. + + Used by :meth:`process_mcp_request` to skip LiteLLM admission auth entirely: the gateway is a + transparent proxy and the caller's ``Authorization`` is an upstream token, never a LiteLLM key. + Mirrors :meth:`_target_servers_delegate_auth_to_upstream`; a mixed-target request keeps normal auth. + """ + from litellm.proxy._experimental.mcp_server.mcp_server_manager import ( + global_mcp_server_manager, + ) + from litellm.types.mcp import MCPAuth + + target_names = MCPRequestHandler._resolve_target_server_names(path=path, mcp_servers_header=mcp_servers) + if not target_names: + return False + + for name in target_names: + server = global_mcp_server_manager.get_mcp_server_by_name(name, client_ip=client_ip) + if server is None or server.auth_type != MCPAuth.true_passthrough: + return False + return True + @staticmethod def _resolve_target_server_names(path: str, mcp_servers_header: Optional[List[str]]) -> List[str]: """ @@ -558,10 +591,19 @@ class MCPRequestHandler: ASGI headers are in format: List[List[bytes, bytes]] We need to convert them to the format Headers expects. + + Collapsing the ASGI list into a dict keeps the last value for a duplicated + header name, so a request carrying more than one ``Authorization`` is + rejected first: for the client-forwarded token modes the gateway relays the + caller's ``Authorization`` upstream, so a duplicate would make which token is + forwarded ambiguous (and diverge from what admission inspected). Multiple + ``Authorization`` headers is malformed for bearer auth anyway (RFC 9110: not + a comma-combinable field), so fail closed with a 400. """ + raw_headers = scope.get("headers", []) + MCPRequestHandler._reject_duplicate_authorization(raw_headers) try: # ASGI headers are list of [name: bytes, value: bytes] pairs - raw_headers = scope.get("headers", []) # Convert bytes to strings and create dict for Headers constructor headers_dict = {name.decode("latin-1"): value.decode("latin-1") for name, value in raw_headers} return Headers(headers_dict) @@ -570,6 +612,26 @@ class MCPRequestHandler: # Return empty Headers object with empty dict return Headers({}) + @staticmethod + def _reject_duplicate_authorization(raw_headers: object) -> None: + """Raise 400 when the raw ASGI headers carry more than one ``Authorization`` header.""" + if not isinstance(raw_headers, (list, tuple)): + return + count = 0 + for entry in raw_headers: + if not isinstance(entry, (list, tuple)) or len(entry) < 1: + continue + name = entry[0] + if isinstance(name, (bytes, bytearray)) and bytes(name).lower() == b"authorization": + count += 1 + elif isinstance(name, str) and name.lower() == "authorization": + count += 1 + if count > 1: + raise HTTPException( + status_code=400, + detail="Multiple Authorization headers are not allowed", + ) + @staticmethod async def get_allowed_mcp_servers( user_api_key_auth: Optional[UserAPIKeyAuth] = None, diff --git a/litellm/proxy/_experimental/mcp_server/db.py b/litellm/proxy/_experimental/mcp_server/db.py index 10081ce19de..e4f8b0c331d 100644 --- a/litellm/proxy/_experimental/mcp_server/db.py +++ b/litellm/proxy/_experimental/mcp_server/db.py @@ -3,7 +3,7 @@ import binascii import hashlib import json from datetime import datetime, timedelta, timezone -from typing import TYPE_CHECKING, Any, Dict, Iterable, List, Optional, Set, Union, cast +from typing import TYPE_CHECKING, Any, Awaitable, Callable, Dict, Iterable, List, Optional, Set, Union, cast from litellm._logging import verbose_proxy_logger from litellm._uuid import uuid @@ -558,7 +558,11 @@ async def delete_mcp_server_from_virtualkey(): pass -async def delete_mcp_server(prisma_client: PrismaClient, server_id: str) -> Optional[LiteLLM_MCPServerTable]: +async def delete_mcp_server( + prisma_client: PrismaClient, + server_id: str, + invalidate_token_cache: Optional[Callable[[str, str], Awaitable[None]]] = None, +) -> Optional[LiteLLM_MCPServerTable]: """ Delete the mcp server from the db by server_id @@ -569,6 +573,12 @@ async def delete_mcp_server(prisma_client: PrismaClient, server_id: str) -> Opti caller-visible error. Each table is cleaned independently so a failure on one still attempts the other. + Each enumerated credential row's user also gets their cached per-user token + invalidated (legacy cache + v2 store, via invalidate_token_cache, defaulting + to the manager's shared invalidation): the caches are keyed by + (user_id, server_id), so without this a re-created server reusing the same + server_id would serve tokens minted for the deleted server until TTL. + Returns the deleted mcp server record if it exists, otherwise None """ deleted_server = await MCPServerRepository(prisma_client).table.delete( @@ -577,6 +587,18 @@ async def delete_mcp_server(prisma_client: PrismaClient, server_id: str) -> Opti }, ) if deleted_server is not None: + credential_user_ids: List[str] = [] + try: + credential_rows = await prisma_client.db.litellm_mcpusercredentials.find_many( + where={"server_id": server_id} + ) + credential_user_ids = [row.user_id for row in credential_rows] + except Exception as e: # noqa: BLE001 - enumeration is best-effort; cached tokens expire by TTL + verbose_proxy_logger.warning( + "MCP server %s deleted but per-user credential enumeration failed; cached tokens expire by TTL: %s", + server_id, + e, + ) for model, label in ( (prisma_client.db.litellm_mcpusercredentials, "credential"), (prisma_client.db.litellm_mcpuserenvvars, "env var"), @@ -591,6 +613,15 @@ async def delete_mcp_server(prisma_client: PrismaClient, server_id: str) -> Opti label, e, ) + if credential_user_ids: + if invalidate_token_cache is None: + from litellm.proxy._experimental.mcp_server.mcp_server_manager import ( + global_mcp_server_manager, + ) + + invalidate_token_cache = global_mcp_server_manager.invalidate_user_oauth_token_cache + for user_id in credential_user_ids: + await invalidate_token_cache(user_id, server_id) return deleted_server @@ -1070,6 +1101,103 @@ async def list_user_oauth_credentials( return results +def _decrypted_credential_field(creds: Dict[str, object], field: str) -> object: + """Return one credential field decrypted with the global salt key; non-string and legacy + plaintext values come back unchanged (decrypt_value_helper returns the original on failure).""" + value = creds.get(field) + if not isinstance(value, str): + return value + return decrypt_value_helper( + value=value, + key=field, + exception_type="debug", + return_original_value=True, + ) + + +def mcp_oauth_token_identity(server: object) -> tuple[object, ...]: + """The upstream-OAuth-token-determining fields of an MCP server: the resource/audience (url, or + spec_path for OpenAPI servers), the OAuth mode/grant (auth_type, oauth2_flow), the + authorization-server endpoints, and the OAuth client + scopes. Mirrors the dashboard's + getOAuthAuthorizationIdentity. When any of these change on a server update, previously stored + per-user tokens were minted for the old identity and are stale. Excludes transport and + delegate_auth_to_upstream, which do not affect what token is minted (RFC 8707/8693). + + client_id/client_secret are compared decrypted: stored values are NaCl-encrypted with a fresh + nonce on every write, so comparing ciphertext would flag every routine save as an identity + change and purge tokens that are still valid.""" + creds = getattr(server, "credentials", None) + if isinstance(creds, str): + try: + parsed: object = json.loads(creds) + except ValueError: + parsed = None + else: + parsed = creds + creds_dict: Dict[str, object] = parsed if isinstance(parsed, dict) else {} + return ( + getattr(server, "url", None), + getattr(server, "spec_path", None), + getattr(server, "auth_type", None), + getattr(server, "oauth2_flow", None), + getattr(server, "authorization_url", None), + getattr(server, "token_url", None), + getattr(server, "registration_url", None), + _decrypted_credential_field(creds_dict, "client_id"), + _decrypted_credential_field(creds_dict, "client_secret"), + creds_dict.get("scopes"), + ) + + +async def purge_user_oauth_credentials_for_server( + prisma_client: PrismaClient, + server_id: str, + invalidate_token_cache: Optional[Callable[[str, str], Awaitable[None]]] = None, +) -> int: + """Delete every stored per-user OAuth token for a server and invalidate each user's cached + token everywhere it can be served from (the legacy per-user token cache and the v2 per-user OAuth + token store), so no user keeps a token minted for a superseded configuration. Called when a server + update changes a mint-relevant field (see mcp_oauth_token_identity). Returns the number of rows + removed. + + LiteLLM_MCPUserCredentials also stores BYOK API keys in the same column; only rows whose payload + decodes as an OAuth2 credential (see _decode_oauth_payload) are deleted, because a config change + only invalidates minted tokens, never a user's own stored key. Rows are therefore deleted per + (user_id, server_id) pair rather than by a blanket server_id filter. An OAuth row inserted while + the purge runs for a user not yet enumerated survives; a re-auth completing in the window for an + already-enumerated user is deleted along with the stale row (the pair delete cannot tell them + apart), which costs that user one extra re-auth and nothing else. + + invalidate_token_cache is injectable for tests; it defaults to the manager's shared + invalidate_user_oauth_token_cache, the single invalidation point for per-user tokens.""" + repo = MCPUserCredentialsRepository(prisma_client) + rows = await repo.table.find_many(where={"server_id": server_id}) + oauth_rows = [row for row in rows if _decode_oauth_payload(row.credential_b64) is not None] + if not oauth_rows: + return 0 + deleted_count = await repo.table.delete_many( + where={"server_id": server_id, "user_id": {"in": [row.user_id for row in oauth_rows]}} + ) + if invalidate_token_cache is None: + from litellm.proxy._experimental.mcp_server.mcp_server_manager import ( + global_mcp_server_manager, + ) + + invalidate_token_cache = global_mcp_server_manager.invalidate_user_oauth_token_cache + + for row in oauth_rows: + await invalidate_token_cache(row.user_id, server_id) + if deleted_count != len(oauth_rows): + verbose_proxy_logger.warning( + "MCP server %s: purge removed %d OAuth credential row(s) but %d were enumerated; " + "row(s) were deleted concurrently during the purge", + server_id, + deleted_count, + len(oauth_rows), + ) + return deleted_count + + async def refresh_user_oauth_token( prisma_client: PrismaClient, user_id: str, diff --git a/litellm/proxy/_experimental/mcp_server/discoverable_endpoints.py b/litellm/proxy/_experimental/mcp_server/discoverable_endpoints.py index fa1f73cea77..4fd47a97066 100644 --- a/litellm/proxy/_experimental/mcp_server/discoverable_endpoints.py +++ b/litellm/proxy/_experimental/mcp_server/discoverable_endpoints.py @@ -465,8 +465,19 @@ async def _store_per_user_token_server_side( def _raise_if_not_oauth2(mcp_server: MCPServer) -> None: - """Reject a non-oauth2 server from the gateway's OAuth authorize/token/register flow.""" - if mcp_server.auth_type == MCPAuth.oauth2: + """Reject a server without upstream OAuth from the gateway's authorize/token/register flow. + + The client-forwarded token modes (``true_passthrough`` / ``oauth_delegate``) are allowed + through: the caller owns the upstream token, and this relayed flow is how a browser obtains + one against the upstream IdP (the admin UI's browser-only Authorize uses it). The minted + token is upstream-audienced and held by the caller; the gateway persists nothing for these + modes (DCR persistence is opt-in and never enabled on this path). + """ + from litellm.proxy._experimental.mcp_server.mcp_server_manager import ( # noqa: PLC0415 # circular import with mcp_server_manager at module load + _UPSTREAM_OAUTH_DISCOVERY_AUTH_TYPES, + ) + + if mcp_server.auth_type in _UPSTREAM_OAUTH_DISCOVERY_AUTH_TYPES: return raise HTTPException( status_code=400, @@ -515,8 +526,7 @@ async def authorize_with_server( response_type: Optional[str] = None, scope: Optional[str] = None, ): - if mcp_server.auth_type != "oauth2": - raise HTTPException(status_code=400, detail="MCP server is not OAuth2") + _raise_if_not_oauth2(mcp_server) if mcp_server.authorization_url is None: raise HTTPException(status_code=400, detail="MCP server authorization url is not set") @@ -581,8 +591,12 @@ async def exchange_token_with_server( if mcp_server.token_url is None: raise HTTPException(status_code=400, detail="MCP server token url is not set") + # The id and secret must come from the same source. When the server-side client_id wins, + # falling back to the caller's secret pairs the persisted client with a foreign secret; the + # register short-circuit hands clients a placeholder secret ("dummy"), so a re-auth against a + # persisted public PKCE client (no stored secret) would send that placeholder and the IdP 401s. resolved_client_id = mcp_server.client_id if mcp_server.client_id else client_id - resolved_client_secret = mcp_server.client_secret if mcp_server.client_secret else client_secret + resolved_client_secret = mcp_server.client_secret if mcp_server.client_id else client_secret try: client_auth = build_token_endpoint_client_auth( auth_method=mcp_server.token_endpoint_auth_method, @@ -1298,11 +1312,15 @@ async def _build_oauth_protected_resource_response( """ Build OAuth protected resource response with the appropriate URL pattern. - For pass-through MCP servers (``MCPServer.is_oauth_passthrough``), the - gateway proxies the upstream's own ``oauth-protected-resource`` metadata - so that standards-compliant MCP clients discover the **upstream** IdP - instead of the gateway. The ``resource`` field is rewritten to the - gateway's own URL so clients present the bearer token back to the gateway. + For pass-through MCP servers, the gateway proxies the upstream's own + ``oauth-protected-resource`` metadata so standards-compliant MCP clients + discover the **upstream** IdP instead of the gateway. For ``true_passthrough`` + and ``oauth_delegate`` the metadata is returned verbatim (``resource`` stays + the upstream): the caller's token is forwarded to and validated by the + upstream, so its audience must be the upstream — rewriting it to the gateway + would make a strict IdP (e.g. Entra) refuse to mint it or the upstream reject + it. Only the legacy ``is_oauth_passthrough`` opt-in rewrites ``resource`` to + the gateway's own URL so clients present the bearer token back to the gateway. Args: request: FastAPI Request object @@ -1343,7 +1361,9 @@ async def _build_oauth_protected_resource_response( # Pass-through branch: proxy the upstream's own metadata so discovery # directs the client at the real IdP (Okta, Keycloak, …) instead of us. - if mcp_server is not None and mcp_server.is_oauth_passthrough: + if mcp_server is not None and ( + mcp_server.is_oauth_passthrough or mcp_server.is_oauth_delegate or mcp_server.is_true_passthrough + ): try: upstream_metadata = await fetch_upstream_oauth_protected_resource(mcp_server) except Exception as exc: @@ -1359,8 +1379,9 @@ async def _build_oauth_protected_resource_response( ) if upstream_metadata is not None: - response = {**upstream_metadata, "resource": resource_url} - return response + if mcp_server.is_true_passthrough or mcp_server.is_oauth_delegate: + return upstream_metadata + return {**upstream_metadata, "resource": resource_url} # Upstream responded but with non-200 or non-dict payload. For # pass-through servers the gateway is NOT the authorization server, diff --git a/litellm/proxy/_experimental/mcp_server/exceptions.py b/litellm/proxy/_experimental/mcp_server/exceptions.py index b3f7ca9bbe2..3e3e549008d 100644 --- a/litellm/proxy/_experimental/mcp_server/exceptions.py +++ b/litellm/proxy/_experimental/mcp_server/exceptions.py @@ -73,3 +73,18 @@ class MCPUpstreamAuthError(Exception): detail=detail, headers={"www-authenticate": challenge} if challenge else None, ) + + +class MCPToolResultError(Exception): + """An MCP tool call completed with ``isError=True`` in its result. + + Never raised on the wire path: streamable HTTP MCP correctly returns tool + failures as HTTP 200 with ``result.isError: true`` per the MCP spec. This + exception only drives the standard failure logging (``status="failure"`` + payload, OTel ERROR span) for such results. + + Lives here rather than ``utils.py`` deliberately: tests reload ``utils`` + to re-read its env-derived constants, and a reload would fork this class + into two identities, breaking ``isinstance`` checks against instances + created before the reload. + """ diff --git a/litellm/proxy/_experimental/mcp_server/mcp_server_manager.py b/litellm/proxy/_experimental/mcp_server/mcp_server_manager.py index 8e4b3c57bbb..8dd9949e17b 100644 --- a/litellm/proxy/_experimental/mcp_server/mcp_server_manager.py +++ b/litellm/proxy/_experimental/mcp_server/mcp_server_manager.py @@ -57,7 +57,11 @@ from litellm.proxy._experimental.mcp_server.elicitation_handler import ( from litellm.proxy._experimental.mcp_server.sampling_handler import ( MCP_SAMPLING_AVAILABLE, ) -from litellm.proxy._experimental.mcp_server.oauth2_token_cache import resolve_mcp_auth +from litellm.proxy._experimental.mcp_server.oauth2_token_cache import ( + MCPPerUserTokenCache, + mcp_per_user_token_cache, + resolve_mcp_auth, +) from litellm.proxy._experimental.mcp_server.outbound_credentials import ( Error, Ok, @@ -81,6 +85,7 @@ from litellm.proxy._experimental.mcp_server.outbound_credentials.token_exchange_ ) from litellm.proxy._experimental.mcp_server.outbound_credentials.types import ( AuthorizationCodeConfig, + PassthroughConfig, ServerSpec, TokenExchangeConfig, ) @@ -170,6 +175,16 @@ _user_env_vars_cache: dict[tuple[str, str], tuple[dict[str, str], float]] = {} _USER_ENV_VARS_CACHE_TTL = 60 # seconds _USER_ENV_VARS_CACHE_MAX_SIZE = 4096 # cap to prevent unbounded growth +# Auth types whose upstream OAuth endpoints (protected-resource + authorization-server metadata) the +# gateway discovers from the upstream itself: interactive oauth2 and the two client-forwarded modes. +# OBO/M2M endpoint discovery is decided separately via _obo_needs_endpoint_discovery. Shared by the +# config-YAML and DB server loaders so the two paths cannot drift on which modes trigger discovery. +_UPSTREAM_OAUTH_DISCOVERY_AUTH_TYPES: tuple[MCPAuth, ...] = ( + MCPAuth.oauth2, + MCPAuth.true_passthrough, + MCPAuth.oauth_delegate, +) + def invalidate_user_env_vars_cache(user_id: str, server_id: str) -> None: """Drop a cached entry after the user stores or clears their env var values @@ -216,6 +231,13 @@ def _should_strip_caller_authorization( pass-through cold-start case (RFC 9728) the bearer in ``Authorization`` is the upstream OAuth token and must be forwarded, so we keep it. + - **oauth_delegate servers**: admission always runs and there is no + anonymous path, so the caller's separate ``Authorization`` is + forwarded only when a distinct ``x-litellm-api-key`` carried + admission. Without that header the ``Authorization`` *was* the + admission credential — a virtual key, an IdP JWT, or an SSO / OIDC / + session token whose ``api_key`` is ``None`` — and must never reach + the upstream, so it is stripped regardless of the ``api_key`` value. """ if mcp_server.auth_type == MCPAuth.oauth2_token_exchange: # OBO: the inbound Authorization is the subject token. It is exchanged at the IdP and only the @@ -229,11 +251,13 @@ def _should_strip_caller_authorization( # upstream — it would override another user's stored credential. Delegate and # pass-through return None from to_server_spec and keep forwarding the bearer. return True - if not mcp_server.is_oauth_passthrough: + if not (mcp_server.is_oauth_passthrough or mcp_server.is_oauth_delegate): return False normalized_raw_headers = {str(k).lower(): v for k, v in (raw_headers or {}).items() if isinstance(k, str)} has_explicit_litellm_admission_header = normalized_raw_headers.get("x-litellm-api-key") is not None + if mcp_server.is_oauth_delegate: + return not has_explicit_litellm_admission_header admission_consumed_authorization_as_litellm_key = ( user_api_key_auth is not None and bool(getattr(user_api_key_auth, "api_key", None)) @@ -326,6 +350,89 @@ async def _resolve_byok_mcp_auth_header( return mcp_auth_header +def _client_forwarded_authorization_headers( + mcp_server: MCPServer, + oauth2_headers: Optional[dict[str, str]], + raw_headers: Optional[dict[str, str]], + user_api_key_auth: Optional[UserAPIKeyAuth], +) -> Optional[dict[str, str]]: + """Egress headers for the client-forwarded-token modes (``true_passthrough`` / ``oauth_delegate``). + + Forwards the caller's ``Authorization`` to the upstream, stripped when + ``_should_strip_caller_authorization`` says it was consumed as the LiteLLM admission key. Shared by + ``_call_regular_mcp_tool`` and ``server.py``'s ``_prepare_mcp_server_headers`` so the two egress + paths cannot drift, mirroring the ``_should_strip_caller_authorization`` split. + """ + extra_headers = oauth2_headers.copy() if oauth2_headers else None + if extra_headers and _should_strip_caller_authorization( + mcp_server=mcp_server, + raw_headers=raw_headers, + user_api_key_auth=user_api_key_auth, + ): + return _without_authorization(extra_headers) + return extra_headers + + +def _take_forwarded_authorization( + headers: Optional[dict[str, str]], +) -> tuple[Optional[str], Optional[dict[str, str]]]: + """Pop the ``Authorization`` value out of ``headers`` (case-insensitive), returning it with the + remaining headers, so the passthrough resolver arm is the single Authorization source rather than + the header also riding in ``extra_headers`` (which the resolved auth would then defer to).""" + if not headers: + return None, headers + value = next((v for k, v in headers.items() if k.lower() == "authorization"), None) + return value, _without_authorization(headers) + + +def _passthrough_token_from_mcp_auth_header( + mcp_auth_header: Optional[Union[str, dict[str, str]]], +) -> Optional[str]: + """The caller's per-server upstream credential for a passthrough-mode server, or None. + + Sourced from ``x-mcp-{alias}-authorization`` (string or per-header dict form) or the deprecated + global ``x-mcp-auth`` fallback. Per-server headers are the multi-server shape: they bind one + token to one server, so an aggregate scope with several passthrough-mode servers never replays + a single credential across upstreams. The value is forwarded verbatim, so it must be the full + header value (e.g. ``Bearer ``).""" + if isinstance(mcp_auth_header, str): + return mcp_auth_header or None + if isinstance(mcp_auth_header, dict): + return next((v for k, v in mcp_auth_header.items() if k.lower() == "authorization"), None) + return None + + +def _consumes_caller_authorization(server: MCPServer) -> bool: + """True when this server's egress forwards the caller's request-wide ``Authorization`` upstream: + the client-forwarded token modes, legacy OAuth pass-through, and legacy upstream-delegated + interactive oauth2. An unstamped oauth2 row (flow column not yet backfilled) reads as a consumer, + which errs toward suppression — the fail-safe direction.""" + if server.is_true_passthrough or server.is_oauth_delegate or server.is_oauth_passthrough: + return True + return ( + server.auth_type == MCPAuth.oauth2 + and getattr(server, "delegate_auth_to_upstream", False) is True + and not server.has_client_credentials + ) + + +def _caller_authorization_fans_out( + server: MCPServer, + scope_servers: Optional[list[MCPServer]], +) -> bool: + """True when forwarding the caller's request-wide ``Authorization`` to ``server`` inside a + listing fan-out would replay one credential against multiple upstreams: another server in the + scope also consumes it (RFC 9700 cross-resource replay). ``scope_servers`` is None for + explicitly-addressed operations (tool call, get_prompt, read_resource, single-server routes), + where the client named the one target and the gateway is not choosing recipients.""" + if scope_servers is None: + return False + return any( + other is not None and other.server_id != server.server_id and _consumes_caller_authorization(other) + for other in scope_servers + ) + + def _extract_upstream_auth_failure( exc: BaseException, ) -> Optional[tuple[int, Optional[str]]]: @@ -696,10 +803,12 @@ class MCPServerManager: self, cred_provider: Optional[UpstreamCredentialProvider] = None, per_user_oauth_token_store: Optional[InvalidatableOAuthTokenStore] = None, + per_user_token_cache: Optional[MCPPerUserTokenCache] = None, ): self._per_user_oauth_token_store = per_user_oauth_token_store or LazyPerUserOAuthTokenStore( self.get_mcp_server_by_id ) + self._per_user_token_cache = per_user_token_cache or mcp_per_user_token_cache self._cred_provider = cred_provider or UpstreamCredentialProvider( oauth_token_store=self._per_user_oauth_token_store, token_exchanger=build_token_exchanger(), @@ -891,7 +1000,7 @@ class MCPServerManager: auth_type = server_config.get("auth_type", None) if server_url and ( - auth_type == MCPAuth.oauth2 + auth_type in _UPSTREAM_OAUTH_DISCOVERY_AUTH_TYPES or self._obo_needs_endpoint_discovery( auth_type, server_config.get("token_exchange_endpoint"), @@ -900,7 +1009,7 @@ class MCPServerManager: ): mcp_oauth_metadata = await self._descovery_metadata( server_url=server_url, - allow_origin_fallback=auth_type == MCPAuth.oauth2, + allow_origin_fallback=auth_type in _UPSTREAM_OAUTH_DISCOVERY_AUTH_TYPES, ) else: mcp_oauth_metadata = None @@ -1292,7 +1401,7 @@ class MCPServerManager: auth_type = cast(MCPAuthType, mcp_server.auth_type) server_url = mcp_server.url needs_discovery = bool(server_url) and ( - (auth_type == MCPAuth.oauth2 and not mcp_server.authorization_url) + (auth_type in _UPSTREAM_OAUTH_DISCOVERY_AUTH_TYPES and not mcp_server.authorization_url) or self._obo_needs_endpoint_discovery( auth_type, mcp_server.token_exchange_endpoint @@ -1303,7 +1412,7 @@ class MCPServerManager: mcp_oauth_metadata = ( await self._descovery_metadata( server_url=server_url, # type: ignore[arg-type] - allow_origin_fallback=auth_type == MCPAuth.oauth2, + allow_origin_fallback=auth_type in _UPSTREAM_OAUTH_DISCOVERY_AUTH_TYPES, ) if needs_discovery else None @@ -1635,15 +1744,18 @@ class MCPServerManager: delegate_server_ids = [ server.server_id for server in self.get_registry().values() - if getattr(server, "auth_type", None) == MCPAuth.oauth2 - and getattr(server, "delegate_auth_to_upstream", False) is True - # M2M servers must not be exposed anonymously: an - # unauthenticated caller would get LiteLLM to proxy tool - # calls using its stored client_credentials. Resolve the flow - # rather than reading has_client_credentials so an unstamped - # M2M-shape row (null column, verbatim-read as non-M2M) still - # fails closed here, matching the anonymous-delegate auth gate. - and MCPServerManager.effective_oauth2_flow(server) != "client_credentials" + if ( + getattr(server, "auth_type", None) == MCPAuth.oauth2 + and getattr(server, "delegate_auth_to_upstream", False) is True + # M2M servers must not be exposed anonymously: an + # unauthenticated caller would get LiteLLM to proxy tool + # calls using its stored client_credentials. Resolve the flow + # rather than reading has_client_credentials so an unstamped + # M2M-shape row (null column, verbatim-read as non-M2M) still + # fails closed here, matching the anonymous-delegate auth gate. + and MCPServerManager.effective_oauth2_flow(server) != "client_credentials" + ) + or getattr(server, "auth_type", None) == MCPAuth.true_passthrough ] combined_servers.update(delegate_server_ids) @@ -2241,16 +2353,17 @@ class MCPServerManager: spec = None if transport == MCPTransport.stdio else to_server_spec(server) provider = cred_provider or self._cred_provider # A caller-supplied per-request override (mcp_auth_header / x-mcp-*) defers to the v1 path - # so it wins - except for the per-user modes the v2 resolver owns (authorization_code's - # stored token and token_exchange's RFC 8693 minted token). A caller must not be able to - # substitute another user's stored credential, nor silently disable the OBO exchange and - # forward an arbitrary bearer upstream, so we keep the v2 spec and ignore the override for - # both; the REST tools preview supplies its not-yet-persisted token through the resolver - # (cred_provider), never this path. + # so it wins - except for the modes the v2 resolver owns per-caller (authorization_code's + # stored token, token_exchange's RFC 8693 minted token, and the passthrough modes' + # forwarded caller token). A caller must not be able to substitute another user's stored + # credential, nor silently disable the OBO exchange and forward an arbitrary bearer + # upstream, so we keep the v2 spec and ignore the override for these; the REST tools + # preview supplies its not-yet-persisted token through the resolver (cred_provider), + # never this path. if ( spec is not None and mcp_auth_header - and not isinstance(spec.config, (AuthorizationCodeConfig, TokenExchangeConfig)) + and not isinstance(spec.config, (AuthorizationCodeConfig, PassthroughConfig, TokenExchangeConfig)) ): spec = None auth_value = ( @@ -2317,11 +2430,17 @@ class MCPServerManager: server_url = server.url or "" if spec is not None: + inbound_token = subject_token + if isinstance(spec.config, PassthroughConfig): + inbound_token, extra_headers = _take_forwarded_authorization(extra_headers) + per_server_token = _passthrough_token_from_mcp_auth_header(mcp_auth_header) + if per_server_token is not None: + inbound_token = per_server_token resolved_auth, extra_headers = await self._resolve_v2_auth( server=server, spec=spec, provider=provider, - subject_token=subject_token, + subject_token=inbound_token, user_api_key_auth=user_api_key_auth, extra_headers=extra_headers, ) @@ -3743,6 +3862,13 @@ class MCPServerManager: user_api_key_auth=user_api_key_auth, ): extra_headers = _without_authorization(extra_headers) + elif mcp_server.is_true_passthrough or mcp_server.is_oauth_delegate: + extra_headers = _client_forwarded_authorization_headers( + mcp_server=mcp_server, + oauth2_headers=oauth2_headers, + raw_headers=raw_headers, + user_api_key_auth=user_api_key_auth, + ) if mcp_server.extra_headers and raw_headers: if extra_headers is None: @@ -3933,10 +4059,13 @@ class MCPServerManager: return await self._cred_provider.has_user_token(to_subject(user_api_key_auth, None), spec) async def invalidate_user_oauth_token_cache(self, user_id: str, server_id: str) -> None: - """Drop the v2 chain's cached token for ``(user_id, server_id)`` after the credential row - changes (re-auth, revoke), so the next resolve reads the new row instead of serving the - replaced token until its cache TTL. Best-effort: a cache-drop failure is logged, never - raised, because the DB write already succeeded and the TTL remains the backstop. + """Drop every cached token for ``(user_id, server_id)`` after the credential row changes + (re-auth, revoke, config-change purge): the v2 chain's cache and the legacy per-user token + cache, so the next resolve reads the new row instead of serving the replaced token until its + cache TTL, whichever path resolves it. This is the single invalidation point for per-user + OAuth tokens; callers must not evict individual caches directly. Best-effort: a cache-drop + failure is logged, never raised, because the DB write already succeeded and the TTL remains + the backstop. """ try: await self._per_user_oauth_token_store.invalidate(user_id, server_id) @@ -3944,6 +4073,12 @@ class MCPServerManager: verbose_logger.warning( "Failed to invalidate cached MCP OAuth token for user=%s server=%s: %s", user_id, server_id, exc ) + try: + await self._per_user_token_cache.delete(user_id, server_id) + except Exception as exc: # noqa: BLE001 - cache drop is best-effort; TTL is the backstop + verbose_logger.warning( + "Failed to drop legacy cached MCP OAuth token for user=%s server=%s: %s", user_id, server_id, exc + ) async def _resolve_oauth2_headers_for_tool_call( self, diff --git a/litellm/proxy/_experimental/mcp_server/outbound_credentials/adapter.py b/litellm/proxy/_experimental/mcp_server/outbound_credentials/adapter.py index 05896bfff74..e87e8081ced 100644 --- a/litellm/proxy/_experimental/mcp_server/outbound_credentials/adapter.py +++ b/litellm/proxy/_experimental/mcp_server/outbound_credentials/adapter.py @@ -23,6 +23,7 @@ from litellm.proxy._experimental.mcp_server.outbound_credentials.types import ( AuthorizationCodeConfig, CredError, NoneConfig, + PassthroughConfig, ServerSpec, SharedKey, Subject, @@ -62,9 +63,10 @@ def to_server_spec(server: MCPServer) -> Optional[ServerSpec]: an ``assert_never`` tail, so a newly added auth mode fails the type gate here until it is explicitly mapped or explicitly deferred, rather than silently falling through to v1. Live modes: ``none``, the static-header family (``api_key`` plus the Authorization schemes, - all shared-key), ``oauth2`` per-user tokens (``authorization_code``), and - ``oauth2_token_exchange`` (OBO); client_credentials (M2M), delegated/passthrough - oauth2, and SigV4 return None and stay on v1. + all shared-key), ``oauth2`` per-user tokens (``authorization_code``), ``oauth2_token_exchange`` + (OBO), and the client-forwarded token modes ``true_passthrough`` / ``oauth_delegate`` + (``PassthroughConfig``); client_credentials (M2M), delegated/passthrough oauth2, and SigV4 + return None and stay on v1. """ if server.is_byok: return None # per-user BYOK source not migrated yet -> defer to v1 (any auth_type) @@ -94,6 +96,8 @@ def to_server_spec(server: MCPServer) -> Optional[ServerSpec]: ) # client_credentials (M2M) and delegate/passthrough oauth2 stay on v1 return None + case MCPAuth.true_passthrough | MCPAuth.oauth_delegate: + return ServerSpec(server_id=server.server_id, resource=resource, config=PassthroughConfig()) case MCPAuth.oauth2_token_exchange: return _token_exchange_spec(server, resource) case MCPAuth.aws_sigv4: diff --git a/litellm/proxy/_experimental/mcp_server/outbound_credentials/resolver.py b/litellm/proxy/_experimental/mcp_server/outbound_credentials/resolver.py index c82ce1037d6..ecfd471190c 100644 --- a/litellm/proxy/_experimental/mcp_server/outbound_credentials/resolver.py +++ b/litellm/proxy/_experimental/mcp_server/outbound_credentials/resolver.py @@ -7,10 +7,11 @@ no precedence cascade. It is wildcard-free with an `assert_never` tail, so addin an arm fails the type gate (basedpyright `reportMatchNotExhaustive`); a bypassed gate fails loudly at runtime instead of returning `None`. -`none` and `api_key` (shared-key source) are live, as is `authorization_code`, which reads the -user's token from the injected `OAuthTokenStore`, and `token_exchange`, which swaps the caller's -inbound token through the injected `TokenExchanger`. The remaining arms are `not_implemented` stubs -that each land in a follow-up PR with their seam. Pure v2: no imports from v1. +`none`, `api_key` (shared-key source), and `passthrough` (forwards the caller's own inbound token) +are live, as is `authorization_code`, which reads the user's token from the injected +`OAuthTokenStore`, and `token_exchange`, which swaps the caller's inbound token through the +injected `TokenExchanger`. The remaining arms are `not_implemented` stubs that each land in a +follow-up PR with their seam. Pure v2: no imports from v1. """ from __future__ import annotations @@ -97,7 +98,7 @@ class UpstreamCredentialProvider: case ApiKeyConfig() as config: return self._api_key(config) case PassthroughConfig(): - return _not_implemented(AuthSpecKind.passthrough) + return self._passthrough(subject) case ClientCredentialsConfig(): return _not_implemented(AuthSpecKind.client_credentials) case TokenExchangeConfig() as config: @@ -118,6 +119,18 @@ class UpstreamCredentialProvider: """ return await self._authz_token(subject, server) is not None + def _passthrough(self, subject: Subject) -> Result[httpx.Auth, CredError]: + """Forward the caller's own upstream credential verbatim; the gateway mints nothing. + + The inbound token is the caller's already-disambiguated ``Authorization`` (never the LiteLLM + admission credential; the edge adapter drops that before building the ``Subject``). When it is + absent the request is sent unauthenticated so the upstream's own 401 surfaces, rather than the + gateway challenging on the upstream's behalf. + """ + if subject.inbound_token is None: + return Ok(NoOpAuth()) + return Ok(StaticHeaderAuth(subject.inbound_token.get_secret_value(), header_name="Authorization")) + def _api_key(self, config: ApiKeyConfig) -> Result[httpx.Auth, CredError]: match config.key_source: case SharedKey() as source: diff --git a/litellm/proxy/_experimental/mcp_server/rest_endpoints.py b/litellm/proxy/_experimental/mcp_server/rest_endpoints.py index d482e537c5d..74f0b488d20 100644 --- a/litellm/proxy/_experimental/mcp_server/rest_endpoints.py +++ b/litellm/proxy/_experimental/mcp_server/rest_endpoints.py @@ -8,6 +8,7 @@ from typing import ( Dict, List, Literal, + Mapping, Optional, Set, Tuple, @@ -68,6 +69,7 @@ if MCP_AVAILABLE: from mcp.types import Tool as MCPTool from litellm.proxy._experimental.mcp_server.mcp_server_manager import ( + _UPSTREAM_OAUTH_DISCOVERY_AUTH_TYPES, global_mcp_server_manager, ) from litellm.proxy._experimental.mcp_server.oauth_utils import ( @@ -78,7 +80,7 @@ if MCP_AVAILABLE: MCPInfo, MCPServer, _apply_toolset_scope, - _fire_mcp_success_logging, + _fire_mcp_tool_call_logging, _tool_name_matches, execute_mcp_tool, filter_tools_by_allowed_tools, @@ -86,23 +88,32 @@ if MCP_AVAILABLE: ######################################################## ############ MCP Server REST API Routes ################# - async def _safe_fire_mcp_success_logging( + async def _safe_fire_mcp_tool_call_logging( logging_obj: Optional[Any], result: Any, start_time: datetime, end_time: datetime, + user_api_key_auth: Optional[UserAPIKeyAuth] = None, + request_data: Optional[Mapping[str, object]] = None, ) -> None: if logging_obj is None: return logging_results = await asyncio.gather( - _fire_mcp_success_logging(logging_obj, result, start_time, end_time), + _fire_mcp_tool_call_logging( + logging_obj, + result, + start_time, + end_time, + user_api_key_auth=user_api_key_auth, + request_data=request_data, + ), return_exceptions=True, ) logging_error = logging_results[0] if isinstance(logging_error, asyncio.CancelledError): raise logging_error if isinstance(logging_error, BaseException): - verbose_logger.warning("MCP tool success logging failed (continuing): %s", logging_error) + verbose_logger.warning("MCP tool call logging failed (continuing): %s", logging_error) def _get_server_auth_header( server, @@ -872,7 +883,14 @@ if MCP_AVAILABLE: raw_headers=virtual_raw_headers, litellm_logging_obj=virtual_logging_obj, ) - await _safe_fire_mcp_success_logging(virtual_logging_obj, result, _tool_start_time, datetime.now()) + await _safe_fire_mcp_tool_call_logging( + virtual_logging_obj, + result, + _tool_start_time, + datetime.now(), + user_api_key_auth=user_api_key_dict, + request_data=data, + ) return result # Validate required parameters early @@ -955,7 +973,14 @@ if MCP_AVAILABLE: litellm_logging_obj=data.get("litellm_logging_obj"), requested_server_id=canonical_server_id, ) - await _safe_fire_mcp_success_logging(logging_obj, result, _tool_start_time, datetime.now()) + await _safe_fire_mcp_tool_call_logging( + logging_obj, + result, + _tool_start_time, + datetime.now(), + user_api_key_auth=user_api_key_dict, + request_data=data, + ) return result except MCPMissingUserEnvVarsError as e: verbose_logger.info( @@ -1297,8 +1322,13 @@ if MCP_AVAILABLE: if isinstance(credentials, dict): mcp_auth_header = credentials.get("auth_value") + # Authorization doubles as the admission fallback (LITELLM_API_KEY_HEADER_NAME_SECONDARY): + # when the primary x-litellm-api-key header is absent, the Authorization value is the + # caller's LiteLLM key, not an upstream token, and must never be forwarded upstream. oauth2_headers: Optional[Dict[str, str]] = None - if new_mcp_server_request.auth_type == MCPAuth.oauth2: + if new_mcp_server_request.auth_type in _UPSTREAM_OAUTH_DISCOVERY_AUTH_TYPES and headers.get( + MCPRequestHandler.LITELLM_API_KEY_HEADER_NAME_PRIMARY + ): oauth2_headers = MCPRequestHandler._get_oauth2_headers_from_headers(headers) async def _list_tools_operation(client): diff --git a/litellm/proxy/_experimental/mcp_server/server.py b/litellm/proxy/_experimental/mcp_server/server.py index fc847182a60..3550237dd65 100644 --- a/litellm/proxy/_experimental/mcp_server/server.py +++ b/litellm/proxy/_experimental/mcp_server/server.py @@ -20,12 +20,14 @@ from typing import ( Callable, Dict, List, + Mapping, Optional, Set, Tuple, Union, cast, ) +from urllib.parse import urlsplit, urlunsplit import httpx from fastapi import FastAPI, HTTPException @@ -47,7 +49,10 @@ from litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp import ( from litellm.proxy._experimental.mcp_server.discoverable_endpoints import ( get_request_base_url, ) -from litellm.proxy._experimental.mcp_server.exceptions import MCPUpstreamAuthError +from litellm.proxy._experimental.mcp_server.exceptions import ( + MCPToolResultError, + MCPUpstreamAuthError, +) from litellm.proxy._experimental.mcp_server.mcp_context import ( _mcp_active_toolset_id, _mcp_gateway_initialize_instructions, @@ -60,6 +65,7 @@ from litellm.proxy._experimental.mcp_server.utils import ( LITELLM_MCP_SERVER_VERSION, MCPMissingUserEnvVarsError, add_server_prefix_to_name, + extract_mcp_tool_result_error_message, get_server_prefix, iter_known_server_prefixes, ) @@ -100,6 +106,27 @@ _MAX_STATEFUL_SESSIONS_PER_OWNER = 100 _MCP_ROUTING_PEEK_MAX_BYTES = 4096 +def _redact_mcp_resource_url(url: Optional[str]) -> Optional[str]: + """Reduce an MCP server URL to its origin (scheme + host + port) for logging. + + Everything else is dropped: userinfo (``user:pass@``), the query string, the + fragment, and the path, because hosted MCP servers routinely embed the + credential in the path (e.g. ``/mcp/s/``) and this value is persisted + in spend-log metadata that a caller who can invoke the tool can read back. + Returns None when the URL has no host to identify (nothing safe to log). + """ + if not isinstance(url, str) or not url: + return None + try: + parts = urlsplit(url) + except ValueError: + return None + if not parts.hostname: + return None + netloc = f"{parts.hostname}:{parts.port}" if parts.port else parts.hostname + return urlunsplit((parts.scheme, netloc, "", "", "")) or None + + def _invalidate_byok_cred_cache(user_id: str, server_id: str) -> None: """Remove a (user_id, server_id) entry from the BYOK credential cache. @@ -331,6 +358,8 @@ if MCP_AVAILABLE: ) from litellm.proxy._experimental.mcp_server.mcp_server_manager import ( MCPServerManager, + _caller_authorization_fans_out, + _client_forwarded_authorization_headers, _should_strip_caller_authorization, _without_authorization, global_mcp_server_manager, @@ -1439,6 +1468,35 @@ if MCP_AVAILABLE: return allowed_mcp_servers + def _client_has_per_server_auth_header( + server: MCPServer, + mcp_server_auth_headers: Optional[Dict[str, Dict[str, str]]], + ) -> bool: + """True if the request carries a per-server ``x-mcp-{alias}-authorization`` + header for this server. This is the multi-server binding: it names one + upstream, so it is unambiguously the caller's upstream token regardless of + auth mode (never the LiteLLM admission credential). + + Resolves through the same ``lookup_mcp_server_auth_in_headers`` egress uses, so + the connect gate and egress agree on which per-server header names match: a + dashboard client sends ``x-mcp-{sanitize_mcp_alias_for_header(alias)}-authorization``, + and matching only the raw alias here would 401 a token egress would forward. + """ + if not mcp_server_auth_headers: + return False + from litellm.proxy._experimental.mcp_server.utils import ( + lookup_mcp_server_auth_in_headers, + ) + + server_headers = lookup_mcp_server_auth_in_headers( + mcp_server_auth_headers, alias=server.alias, server_name=server.server_name + ) + if isinstance(server_headers, str): + return bool(server_headers.strip()) + if isinstance(server_headers, dict): + return any(isinstance(hk, str) and hk.lower() == "authorization" for hk in server_headers) + return False + def _client_has_passthrough_authorization( server: MCPServer, oauth2_headers: Optional[Dict[str, str]], @@ -1456,24 +1514,7 @@ if MCP_AVAILABLE: for k in oauth2_headers.keys(): if k.lower() == "authorization": return True - if mcp_server_auth_headers: - for key in (server.alias, server.server_name, server.name): - if not key: - continue - server_headers = None - for k, v in mcp_server_auth_headers.items(): - if k.lower() == key.lower(): - server_headers = v - break - if server_headers is None: - continue - if isinstance(server_headers, str) and server_headers.strip(): - return True - if isinstance(server_headers, dict): - for hk in server_headers.keys(): - if hk.lower() == "authorization": - return True - return False + return _client_has_per_server_auth_header(server, mcp_server_auth_headers) async def _get_user_oauth_extra_headers_from_db( server: MCPServer, @@ -1528,8 +1569,16 @@ if MCP_AVAILABLE: oauth2_headers: Optional[Dict[str, str]], raw_headers: Optional[Dict[str, str]], user_api_key_auth: Optional[UserAPIKeyAuth] = None, + scope_servers: Optional[list[MCPServer]] = None, ) -> Tuple[Optional[Union[Dict[str, str], str]], Optional[Dict[str, str]]]: - """Build auth and extra headers for a server.""" + """Build auth and extra headers for a server. + + ``scope_servers`` is the full server list a fan-out handler iterates. Passing it lets the + client-forwarded token modes withhold the caller's request-wide ``Authorization`` when + another server in the scope would also receive it (``_caller_authorization_fans_out``); + explicitly-addressed operations leave it None. Per-server ``x-mcp-{alias}-authorization`` + headers are unaffected — they bind one token to one server and are the multi-server shape. + """ server_auth_header: Optional[Union[Dict[str, str], str]] = None if mcp_server_auth_headers: from litellm.proxy._experimental.mcp_server.utils import ( @@ -1543,6 +1592,16 @@ if MCP_AVAILABLE: ) extra_headers: Optional[Dict[str, str]] = None + is_client_forwarded_mode = server.is_true_passthrough or server.is_oauth_delegate + # In a multi-server listing scope the request-wide Authorization can only carry one token, + # so it is withheld from a client-forwarded server when another server in scope also consumes + # it (RFC 9700 cross-resource replay); such scopes must bind per-server via + # x-mcp-{alias}-authorization. The decision is computed once so BOTH the forwarding branch and + # the extra_headers copy loop below honor it — otherwise a server that lists Authorization in + # extra_headers would re-copy the withheld bearer from raw_headers and replay it anyway. + withhold_forwarded_authorization = is_client_forwarded_mode and _caller_authorization_fans_out( + server, scope_servers + ) if server.auth_type == MCPAuth.oauth2: # For OAuth2 M2M servers, upstream Authorization must come from # client_credentials token fetch, never from caller headers. @@ -1561,6 +1620,14 @@ if MCP_AVAILABLE: user_api_key_auth=user_api_key_auth, ): extra_headers = _without_authorization(extra_headers) + elif is_client_forwarded_mode: + if not withhold_forwarded_authorization: + extra_headers = _client_forwarded_authorization_headers( + mcp_server=server, + oauth2_headers=oauth2_headers, + raw_headers=raw_headers, + user_api_key_auth=user_api_key_auth, + ) if server.extra_headers and raw_headers: if extra_headers is None: @@ -1581,7 +1648,9 @@ if MCP_AVAILABLE: for header in server.extra_headers: if not isinstance(header, str): continue - if header.lower() == "authorization" and strip_caller_authorization: + if header.lower() == "authorization" and ( + strip_caller_authorization or withhold_forwarded_authorization + ): continue header_value = normalized_raw_headers.get(header.lower()) if header_value is None: @@ -1785,6 +1854,7 @@ if MCP_AVAILABLE: oauth2_headers=oauth2_headers, raw_headers=raw_headers, user_api_key_auth=user_api_key_auth, + scope_servers=allowed_mcp_servers, ) # Prefer server-stored per-user OAuth when configured, so a stale @@ -1971,6 +2041,7 @@ if MCP_AVAILABLE: oauth2_headers=oauth2_headers, raw_headers=raw_headers, user_api_key_auth=user_api_key_auth, + scope_servers=allowed_mcp_servers, ) try: @@ -2023,6 +2094,7 @@ if MCP_AVAILABLE: oauth2_headers=oauth2_headers, raw_headers=raw_headers, user_api_key_auth=user_api_key_auth, + scope_servers=allowed_mcp_servers, ) try: @@ -2073,6 +2145,7 @@ if MCP_AVAILABLE: oauth2_headers=oauth2_headers, raw_headers=raw_headers, user_api_key_auth=user_api_key_auth, + scope_servers=allowed_mcp_servers, ) try: @@ -2743,12 +2816,40 @@ if MCP_AVAILABLE: return response - async def _fire_mcp_success_logging( + _MCP_CREDENTIAL_REQUEST_FIELDS = frozenset( + { + "raw_headers", + "mcp_auth_header", + "mcp_server_auth_headers", + "oauth2_headers", + "user_api_key_auth", + } + ) + + async def _fire_mcp_tool_call_logging( logging_obj: LiteLLMLoggingObj, result: Any, start_time: datetime, end_time: datetime, + user_api_key_auth: Optional[UserAPIKeyAuth] = None, + request_data: Optional[Mapping[str, object]] = None, ) -> None: + """Fire post-call logging for an executed MCP tool call. + + A result with ``isError=True`` is logged as a failure (``status="failure"`` + payload, so OTel marks the span ERROR) while the HTTP wire behavior stays + 200 + ``isError: true`` per the MCP spec. The error check runs after + ``async_post_mcp_tool_call_hook`` because guardrails may flip the result + to ``isError=True`` in that hook. Raised exceptions never reach here (the + ``@client`` wrapper and ``call_mcp_tool``'s except path log those), so + this cannot double-log a failure. + + ``request_data`` may carry credential-bearing fields (the REST path puts + ``raw_headers``, ``mcp_auth_header``, ``mcp_server_auth_headers``, and + ``oauth2_headers`` at the top level of its data dict), so those are + stripped before the dict is handed to ``post_call_failure_hook`` + callbacks. + """ logging_obj.post_call(original_response=result) await logging_obj.async_post_mcp_tool_call_hook( kwargs=logging_obj.model_call_details, @@ -2757,7 +2858,31 @@ if MCP_AVAILABLE: end_time=end_time, ) logging_obj.call_type = CallTypes.call_mcp_tool.value - await logging_obj.async_success_handler(result=result, start_time=start_time, end_time=end_time) + error_message = extract_mcp_tool_result_error_message(result) + if error_message is None: + await logging_obj.async_success_handler(result=result, start_time=start_time, end_time=end_time) + return + + logging_obj.has_run_logging(event_type="sync_success") + logging_obj.has_run_logging(event_type="async_success") + tool_error = MCPToolResultError(error_message) + logging_obj.failure_handler(tool_error, "", start_time, end_time) + await logging_obj.async_failure_handler(tool_error, "", start_time, end_time) + + if user_api_key_auth is None: + return + from litellm.proxy.proxy_server import proxy_logging_obj + + if proxy_logging_obj: + sanitized_request_data = { + key: value for key, value in (request_data or {}).items() if key not in _MCP_CREDENTIAL_REQUEST_FIELDS + } + await proxy_logging_obj.post_call_failure_hook( + request_data=sanitized_request_data, + original_exception=tool_error, + user_api_key_dict=user_api_key_auth, + route="/mcp/call_tool", + ) @client async def call_mcp_tool( @@ -2833,7 +2958,14 @@ if MCP_AVAILABLE: raise if litellm_logging_obj: - await _fire_mcp_success_logging(litellm_logging_obj, response, start_time, datetime.now()) + await _fire_mcp_tool_call_logging( + logging_obj=litellm_logging_obj, + result=response, + start_time=start_time, + end_time=datetime.now(), + user_api_key_auth=user_api_key_auth, + request_data=kwargs, + ) return response async def mcp_get_prompt( @@ -2954,6 +3086,8 @@ if MCP_AVAILABLE: mcp_server_logo_url=mcp_info.get("logo_url"), namespaced_tool_name=namespaced_tool_name, mcp_session_id=session_id, + mcp_auth_mode=mcp_server.auth_type, + mcp_server_resource=_redact_mcp_resource_url(mcp_server.url), ) else: return StandardLoggingMCPToolCall( @@ -3518,6 +3652,41 @@ if MCP_AVAILABLE: headers={"www-authenticate": www_authenticate}, ) + if ( + server + and server.is_oauth_delegate + and len(mcp_servers or []) == 1 + and _get_forwarded_auth_from_scope(scope) is None + and not _client_has_per_server_auth_header(server, mcp_server_auth_headers) + ): + www_authenticate = _get_passthrough_www_authenticate( + scope=scope, + server_name=server_name, + ) + raise HTTPException( + status_code=401, + detail="Unauthorized", + headers={"www-authenticate": www_authenticate}, + ) + + if ( + server + and server.is_true_passthrough + and len(mcp_servers or []) == 1 + and not _scope_has_authorization_header(scope) + and not _client_has_per_server_auth_header(server, mcp_server_auth_headers) + ): + upstream_status, upstream_www_authenticate = await _probe_upstream_auth(server.url or "", "") + if upstream_status == 401 and upstream_www_authenticate: + raise HTTPException( + status_code=401, + detail="Unauthorized", + headers={"www-authenticate": upstream_www_authenticate}, + ) + + def _scope_has_authorization_header(scope: Scope) -> bool: + return any(key.lower() == b"authorization" for key, _ in scope.get("headers", [])) + def _get_forwarded_auth_from_scope(scope: Scope) -> Optional[str]: """Return the upstream-bound ``Authorization`` header value, or None. @@ -3545,7 +3714,7 @@ if MCP_AVAILABLE: url: str, auth_header: str, timeout: float = 5.0, - ) -> tuple: + ) -> tuple[int, Optional[str]]: """JSON-RPC initialize-probe the upstream URL to check whether the token is accepted. Uses POST so StreamableHTTP MCP servers run the same auth path as a @@ -3575,8 +3744,8 @@ if MCP_AVAILABLE: }, } probe_headers = { - "Authorization": auth_header, "Accept": "application/json, text/event-stream", + **({"Authorization": auth_header} if auth_header else {}), } try: resp = await client.post( diff --git a/litellm/proxy/_experimental/mcp_server/utils.py b/litellm/proxy/_experimental/mcp_server/utils.py index c9c60030dbc..80a469b8c1a 100644 --- a/litellm/proxy/_experimental/mcp_server/utils.py +++ b/litellm/proxy/_experimental/mcp_server/utils.py @@ -415,6 +415,25 @@ def validate_mcp_server_name(server_name: str, raise_http_exception: bool = Fals raise Exception(error_message) +def extract_mcp_tool_result_error_message(result: object) -> Optional[str]: + """The first text content of an ``isError=True`` tool result, or ``None`` + when the result is not an error. + + Accepts both ``mcp.types.CallToolResult`` objects and their dict + equivalents, duck-typed so the ``mcp`` package is not required. + """ + is_error: object = result.get("isError") if isinstance(result, Mapping) else getattr(result, "isError", None) + if is_error is not True: + return None + content: object = result.get("content") if isinstance(result, Mapping) else getattr(result, "content", None) + if isinstance(content, (list, tuple)): + for item in content: + text: object = item.get("text") if isinstance(item, Mapping) else getattr(item, "text", None) + if isinstance(text, str) and text: + return text + return "MCP tool call returned isError=true" + + TOOL_DISPLAY_NAME_PATTERN = re.compile(r"^[a-zA-Z0-9_-]+$") diff --git a/litellm/proxy/auth/auth_checks.py b/litellm/proxy/auth/auth_checks.py index e7fee8d6eb2..cd8103abf5e 100644 --- a/litellm/proxy/auth/auth_checks.py +++ b/litellm/proxy/auth/auth_checks.py @@ -98,7 +98,10 @@ from litellm.repositories.user_repository import UserRepository from litellm.router import Router from litellm.utils import get_utc_datetime -from .auth_checks_organization import organization_role_based_access_check +from .auth_checks_organization import ( + add_team_org_context_to_request_body, + organization_role_based_access_check, +) from .auth_utils import get_model_from_request if TYPE_CHECKING: @@ -707,10 +710,28 @@ async def common_checks( # 10 [OPTIONAL] Organization RBAC checks organization_role_based_access_check(user_object=user_object, route=route, request_body=request_body) + async def _fetch_team_org_id(team_id: str) -> Optional[str]: + try: + team = await get_team_object( + team_id=team_id, + prisma_client=prisma_client, + user_api_key_cache=user_api_key_cache, + proxy_logging_obj=proxy_logging_obj, + ) + except HTTPException: + return None + return team.organization_id + + request_body_for_route_check = await add_team_org_context_to_request_body( + route=route, + request_body=request_body, + fetch_team_org_id=_fetch_team_org_id, + ) + _is_route_allowed = _is_api_route_allowed( route=route, request=request, - request_data=request_body, + request_data=request_body_for_route_check, valid_token=valid_token, user_obj=user_object, ) diff --git a/litellm/proxy/auth/auth_checks_organization.py b/litellm/proxy/auth/auth_checks_organization.py index 44c1d158cbe..b4caff9b8ee 100644 --- a/litellm/proxy/auth/auth_checks_organization.py +++ b/litellm/proxy/auth/auth_checks_organization.py @@ -2,7 +2,7 @@ Auth Checks for Organizations """ -from typing import Dict, List, Optional, Tuple +from typing import Awaitable, Callable, Dict, List, Optional, Tuple from fastapi import status @@ -170,3 +170,33 @@ def _user_is_org_admin( # User must be admin of ALL requested orgs, not just any one return all(org_id in admin_org_ids for org_id in candidate_org_ids) + + +TEAM_ORG_CONTEXT_ROUTES = frozenset({"/team/update"}) + + +async def add_team_org_context_to_request_body( + route: str, + request_body: dict, + fetch_team_org_id: Callable[[str], Awaitable[Optional[str]]], +) -> dict: + """ + Return a copy of request_body with organization_id resolved from the target + team when the route identifies the team by team_id and the caller did not + pass organization_id. This lets an org admin of the team's own org reach the + org-scoped branch of the route gate (which keys off organization_id) without + the client having to send it. Returns request_body unchanged when it does + not apply, so callers that already pass organization_id and non-team routes + are untouched. + """ + if route not in TEAM_ORG_CONTEXT_ROUTES: + return request_body + if request_body.get("organization_id"): + return request_body + team_id = request_body.get("team_id") + if not isinstance(team_id, str) or not team_id: + return request_body + org_id = await fetch_team_org_id(team_id) + if not org_id: + return request_body + return {**request_body, "organization_id": org_id} diff --git a/litellm/proxy/guardrails/_content_utils.py b/litellm/proxy/guardrails/_content_utils.py index 1d31e33c77c..6cdf49818e6 100644 --- a/litellm/proxy/guardrails/_content_utils.py +++ b/litellm/proxy/guardrails/_content_utils.py @@ -32,6 +32,9 @@ def is_text_content_call_type(call_type: str) -> bool: return call_type in TEXT_CONTENT_CALL_TYPES +TEXT_PART_TYPES: FrozenSet[str] = frozenset({"text", "input_text", "output_text"}) + + def _iter_text_parts_in_content(content: Any) -> Iterator[str]: """Yield text fragments from a ``message.content`` value (string or multimodal list). Non-text parts (images, audio, …) are skipped.""" @@ -48,7 +51,7 @@ def _iter_text_parts_in_content(content: Any) -> Iterator[str]: continue if not isinstance(part, dict): continue - if part.get("type") == "text": + if part.get("type") in TEXT_PART_TYPES: text = part.get("text") if isinstance(text, str) and text: yield text @@ -58,14 +61,20 @@ def _coerce_input_to_messages(input_value: Any) -> List[Dict[str, Any]]: """Coerce a Responses-API ``data["input"]`` value into chat-style messages.""" if isinstance(input_value, str): return [{"role": "user", "content": input_value}] - if isinstance(input_value, list): - if input_value and all(isinstance(item, dict) and "role" in item for item in input_value): - return list(input_value) - # Mixed lists (content-part dicts + bare strings) and pure - # string/dict lists all become a single user message; the content - # iterator below handles each element type uniformly. - return [{"role": "user", "content": input_value}] - return [] + if not isinstance(input_value, list): + return [] + messages: List[Dict[str, Any]] = [] + for item in input_value: + if isinstance(item, str): + messages.append({"role": "user", "content": item}) + elif isinstance(item, dict): + if item.get("type") in TEXT_PART_TYPES: + messages.append({"role": item.get("role") or "user", "content": [item]}) + elif "content" in item: + messages.append({"role": item.get("role") or "user", "content": item["content"]}) + elif item.get("type") == "function_call_output" and "output" in item: + messages.append({"role": item.get("role") or "tool", "content": item["output"]}) + return messages def _iter_inspection_messages(data: Dict[str, Any]) -> Iterator[Dict[str, Any]]: @@ -112,7 +121,7 @@ def walk_user_text(data: Dict[str, Any], visit: Callable[[str], str]) -> int: new_parts.append(visit(part)) elif ( isinstance(part, dict) - and part.get("type") == "text" + and part.get("type") in TEXT_PART_TYPES and isinstance(part.get("text"), str) and part["text"] ): @@ -136,25 +145,20 @@ def walk_user_text(data: Dict[str, Any], visit: Callable[[str], str]) -> int: data["input"] = visit(input_value) return visited if isinstance(input_value, list): - # List of full messages: rewrite each message's content. - if input_value and all(isinstance(item, dict) and "role" in item for item in input_value): - for item in input_value: - if "content" in item: - item["content"] = _rewrite_content(item["content"]) - return visited - # List of content parts and/or bare strings: rewrite in place. for idx, item in enumerate(input_value): - if isinstance(item, str) and item: - visited += 1 - input_value[idx] = visit(item) - elif ( - isinstance(item, dict) - and item.get("type") == "text" - and isinstance(item.get("text"), str) - and item["text"] - ): - visited += 1 - input_value[idx] = {**item, "text": visit(item["text"])} + if isinstance(item, str): + if item: + visited += 1 + input_value[idx] = visit(item) + elif isinstance(item, dict): + if item.get("type") in TEXT_PART_TYPES: + if isinstance(item.get("text"), str) and item["text"]: + visited += 1 + input_value[idx] = {**item, "text": visit(item["text"])} + elif "content" in item: + item["content"] = _rewrite_content(item["content"]) + elif item.get("type") == "function_call_output" and "output" in item: + item["output"] = _rewrite_content(item["output"]) return visited return visited diff --git a/litellm/proxy/guardrails/guardrail_hooks/aim/aim.py b/litellm/proxy/guardrails/guardrail_hooks/aim/aim.py index e7d9406ae3b..d22243cbe88 100644 --- a/litellm/proxy/guardrails/guardrail_hooks/aim/aim.py +++ b/litellm/proxy/guardrails/guardrail_hooks/aim/aim.py @@ -93,11 +93,10 @@ class AimGuardrail(CustomGuardrail): user_email=user_email, litellm_call_id=call_id, ) - # Covers multimodal list content + Responses-API input. response = await self.async_handler.post( f"{self.api_base}/fw/v1/analyze", headers=headers, - json={"messages": build_inspection_messages(data)}, + json={"messages": self._build_aim_inspection_messages(data)}, ) response.raise_for_status() res = response.json() @@ -116,6 +115,15 @@ class AimGuardrail(CustomGuardrail): verbose_proxy_logger.error(f"Aim: {action_type} action") return data + @staticmethod + def _build_aim_inspection_messages(data: dict) -> list[dict[str, str]]: + """AIM validates against the OpenAI chat schema. Bare ``role: "tool"`` + without ``tool_call_id`` and bare ``role: "function"`` without ``name`` + are rejected; the flatten drops those fields, so any role outside + ``{system, user, assistant}`` collapses to ``user`` for the AIM POST.""" + safe_roles = {"system", "user", "assistant"} + return [{**m, "role": "user"} if m["role"] not in safe_roles else m for m in build_inspection_messages(data)] + @staticmethod def _rejection(message: str, *, openai_code: str | None = None) -> ProxyException: return ProxyException( @@ -177,7 +185,10 @@ class AimGuardrail(CustomGuardrail): user_email=user_email, litellm_call_id=call_id, ), - json={"messages": build_inspection_messages(request_data) + [{"role": "assistant", "content": output}]}, + json={ + "messages": self._build_aim_inspection_messages(request_data) + + [{"role": "assistant", "content": output}] + }, ) response.raise_for_status() res = response.json() diff --git a/litellm/proxy/guardrails/guardrail_hooks/bedrock_guardrails.py b/litellm/proxy/guardrails/guardrail_hooks/bedrock_guardrails.py index 6e46f971dd8..a45719d2eb8 100644 --- a/litellm/proxy/guardrails/guardrail_hooks/bedrock_guardrails.py +++ b/litellm/proxy/guardrails/guardrail_hooks/bedrock_guardrails.py @@ -33,7 +33,7 @@ import litellm from litellm._logging import verbose_proxy_logger from litellm.litellm_core_utils.core_helpers import redact_nested_match_and_regex_keys from litellm.caching import DualCache -from litellm.exceptions import GuardrailInterventionNormalStringError +from litellm.exceptions import ModifyResponseException from litellm.integrations.custom_guardrail import CustomGuardrail from litellm.llms.bedrock.base_aws_llm import BaseAWSLLM from litellm.llms.custom_httpx.http_handler import ( @@ -754,7 +754,9 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM): ) bedrock_guardrail_response = BedrockGuardrailResponse(**_json_response) if self._should_raise_guardrail_blocked_exception(bedrock_guardrail_response): - raise self._get_http_exception_for_blocked_guardrail(bedrock_guardrail_response) + raise self._get_http_exception_for_blocked_guardrail( + bedrock_guardrail_response, request_data=request_data + ) else: status_code, detail_message = self._parse_bedrock_guardrail_error_response(httpx_response) verbose_proxy_logger.error( @@ -1027,8 +1029,8 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM): return blocked def _get_http_exception_for_blocked_guardrail( - self, response: BedrockGuardrailResponse - ) -> Union[HTTPException, GuardrailInterventionNormalStringError]: + self, response: BedrockGuardrailResponse, request_data: Optional[dict] = None + ) -> Union[HTTPException, ModifyResponseException]: """ Get the HTTP exception for a blocked guardrail. """ @@ -1040,7 +1042,13 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM): bedrock_guardrail_output_text += output.get("text") or "" if self.disable_exception_on_block is True: - return GuardrailInterventionNormalStringError(message=bedrock_guardrail_output_text) + _request_data = request_data or {} + return ModifyResponseException( + message=bedrock_guardrail_output_text, + model=_request_data.get("model", "bedrock-guardrail"), + request_data=_request_data, + guardrail_name=self.guardrail_name, + ) detail: Dict[str, Any] = { "error": "Violated guardrail policy", @@ -1134,18 +1142,6 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM): # This means all actions were ANONYMIZED or NONE, so don't raise exception return False - def create_guardrail_blocked_response(self, response: str) -> ModelResponse: - from litellm.types.utils import Choices, Message, ModelResponse - - return ModelResponse( - choices=[ - Choices( - message=Message(content=response), - ) - ], - model="bedrock-guardrail", - ) - async def async_pre_call_hook( self, user_api_key_dict: UserAPIKeyAuth, @@ -1183,16 +1179,15 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM): ######################################################### ########## 1. Make the Bedrock API request ########## ######################################################### - bedrock_guardrail_response: Optional[Union[BedrockGuardrailResponse, str]] = None - try: - bedrock_guardrail_response = await self.make_bedrock_api_request( - source="INPUT", - messages=filtered_messages, - request_data=data, - logging_event_type=GuardrailEventHooks.pre_call, - ) - except GuardrailInterventionNormalStringError as e: - bedrock_guardrail_response = e.message + # A block with disable_exception_on_block=True raises ModifyResponseException + # from make_bedrock_api_request; that propagates to the endpoint handler, + # which returns a 200 whose message is the guardrail's blockedInputMessaging. + bedrock_guardrail_response = await self.make_bedrock_api_request( + source="INPUT", + messages=filtered_messages, + request_data=data, + logging_event_type=GuardrailEventHooks.pre_call, + ) ######################################################### ######################################################### @@ -1207,8 +1202,6 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM): updated_target_messages=updated_subset, target_indices=filter_result.target_indices, ) - if isinstance(bedrock_guardrail_response, str): - data["mock_response"] = self.create_guardrail_blocked_response(response=bedrock_guardrail_response) ######################################################### ########## 3. Add the guardrail to the applied guardrails header ########## @@ -1248,16 +1241,19 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM): ######################################################### ########## 1. Make the Bedrock API request ########## ######################################################### - bedrock_guardrail_response: Optional[Union[BedrockGuardrailResponse, str]] = None - try: - bedrock_guardrail_response = await self.make_bedrock_api_request( - source="INPUT", - messages=filtered_messages, - request_data=data, - logging_event_type=GuardrailEventHooks.during_call, - ) - except GuardrailInterventionNormalStringError as e: - bedrock_guardrail_response = e.message + # A block with disable_exception_on_block=True raises ModifyResponseException + # from make_bedrock_api_request. Because during_call runs in an asyncio.gather + # alongside the LLM call (common_request_processing.py), swallowing the + # exception here to set data["mock_response"] was ineffective: route_request + # unpacked kwargs before this hook ran, and the LLM task's response was taken + # unconditionally. Letting the exception propagate cancels the LLM task and + # the endpoint handler returns the block response. + bedrock_guardrail_response = await self.make_bedrock_api_request( + source="INPUT", + messages=filtered_messages, + request_data=data, + logging_event_type=GuardrailEventHooks.during_call, + ) ######################################################### ######################################################### @@ -1272,8 +1268,6 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM): updated_target_messages=updated_subset, target_indices=filter_result.target_indices, ) - if isinstance(bedrock_guardrail_response, str): - data["mock_response"] = self.create_guardrail_blocked_response(response=bedrock_guardrail_response) ######################################################### ########## 3. Add the guardrail to the applied guardrails header ########## @@ -1323,7 +1317,11 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM): # users should configure if they want input validation. Running an # extra INPUT scan here produced a duplicate post-call entry in the # trace and made no semantic sense for a "post-call" event. - output_content_bedrock: Optional[Union[BedrockGuardrailResponse, str]] = None + # A block with disable_exception_on_block=True raises ModifyResponseException + # from make_bedrock_api_request; that propagates to the endpoint handler, + # which returns a 200 whose message is the guardrail's blockedInputMessaging. + # Attach the LLM response to original_response so the synthetic block reply + # reports the real token usage the upstream call consumed instead of zero. try: output_content_bedrock = await self.make_bedrock_api_request( source="OUTPUT", @@ -1332,15 +1330,15 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM): request_data=data, logging_event_type=GuardrailEventHooks.post_call, ) - except GuardrailInterventionNormalStringError as e: - output_content_bedrock = e.message + except ModifyResponseException as e: + if e.original_response is None: + e.original_response = response + raise ######################################################### ########## 2. Apply masking to response with output guardrail response ########## ######################################################### - if isinstance(output_content_bedrock, str): - response = self.create_guardrail_blocked_response(response=output_content_bedrock) - elif output_content_bedrock is not None: + if output_content_bedrock is not None: self._apply_masking_to_response( response=response, bedrock_guardrail_response=output_content_bedrock, @@ -1357,7 +1355,7 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM): def _update_messages_with_updated_bedrock_guardrail_response( self, messages: List[AllMessageValues], - bedrock_guardrail_response: Union[BedrockGuardrailResponse, str], + bedrock_guardrail_response: BedrockGuardrailResponse, ) -> List[AllMessageValues]: """ Use the output from the bedrock guardrail to mask sensitive content in messages. @@ -1369,8 +1367,6 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM): Returns: List of messages with content masked according to guardrail response """ - if isinstance(bedrock_guardrail_response, str): - return messages # Get masked texts from guardrail response masked_texts = self._extract_masked_texts_from_response(bedrock_guardrail_response) @@ -1422,7 +1418,14 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM): # pre_call / during_call. Bedrock will raise if the response # violates the guardrail policy. ################################################################### - output_guardrail_response: Optional[Union[BedrockGuardrailResponse, str]] = None + # A block with disable_exception_on_block=True raises ModifyResponseException + # from make_bedrock_api_request. Non-streaming paths let it propagate so + # the endpoint handler turns it into a 200. Streaming can't do that: the + # SSE response headers are already flushed, so a raise would be serialized + # as an error frame by async_streaming_data_generator. Instead, replace + # the assembled response with the synthetic block content in-place and + # yield it as a normal stream, matching the shape a non-streaming block + # produces. try: output_guardrail_response = await self.make_bedrock_api_request( source="OUTPUT", @@ -1431,15 +1434,31 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM): request_data=request_data, logging_event_type=GuardrailEventHooks.post_call, ) - except GuardrailInterventionNormalStringError as e: - output_guardrail_response = e.message + except ModifyResponseException as e: + # Preserve upstream usage from the LLM call we already + # consumed. Non-streaming blocks carry it via + # ModifyResponseException.original_response + + # _blocked_response_usage; streaming has to do the copy + # itself since the exception can't escape this generator. + _original_usage = getattr(assembled_model_response, "usage", None) + assembled_model_response = ModelResponse( + choices=[ + Choices( + index=0, + message=Message(role="assistant", content=e.message), + finish_reason="content_filter", + ) + ], + model=e.model, + ) + if _original_usage is not None: + assembled_model_response.usage = _original_usage + output_guardrail_response = None ######################################################################### ########## 2. Apply masking to response with output guardrail response ########## ######################################################################### - if isinstance(output_guardrail_response, str): - assembled_model_response = self.create_guardrail_blocked_response(response=output_guardrail_response) - elif output_guardrail_response is not None: + if output_guardrail_response is not None: self._apply_masking_to_response( response=assembled_model_response, bedrock_guardrail_response=output_guardrail_response, @@ -1732,13 +1751,13 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM): inputs["texts"] = masked_texts return inputs - except (HTTPException, GuardrailInterventionNormalStringError): - # Let guardrail blocking exceptions propagate as-is so the proxy - # can return the correct HTTP status (400) or handle the - # GuardrailInterventionNormalStringError for disable_exception_on_block mode. - # Without this, the generic except below wraps them into a plain - # Exception, losing the HTTP semantics and preventing the proxy - # from properly blocking the call. + except (HTTPException, ModifyResponseException): + # Let guardrail blocking exceptions propagate as-is so the proxy can + # return the correct HTTP status (400 for HTTPException, 200 with the + # block message for ModifyResponseException in disable_exception_on_block + # mode). Without this, the generic except below wraps them into a plain + # Exception, losing the semantics and preventing the proxy from + # properly blocking the call. raise except Exception as e: verbose_proxy_logger.error("Bedrock Guardrail: Failed to apply guardrail: %s", str(e)) diff --git a/litellm/proxy/guardrails/guardrail_hooks/grayswan/grayswan.py b/litellm/proxy/guardrails/guardrail_hooks/grayswan/grayswan.py index a14d2fc8608..9805b1a9117 100644 --- a/litellm/proxy/guardrails/guardrail_hooks/grayswan/grayswan.py +++ b/litellm/proxy/guardrails/guardrail_hooks/grayswan/grayswan.py @@ -213,7 +213,7 @@ class GraySwanGuardrail(CustomGuardrail): verbose_proxy_logger.debug("Gray Swan Guardrail: dynamic extra_body=%s", safe_dumps(dynamic_body)) # Prepare and send payload - payload = self._prepare_payload(messages, dynamic_body, request_data) + payload = self._prepare_payload(messages, dynamic_body, request_data, logging_obj) if payload is None: return inputs @@ -502,10 +502,38 @@ class GraySwanGuardrail(CustomGuardrail): "grayswan-api-key": self.api_key, } + def _extract_inbound_headers( + self, + request_data: dict, + logging_obj: Optional["LiteLLMLoggingObj"] = None, + ) -> Optional[dict[str, str]]: + headers = (request_data.get("proxy_server_request") or {}).get("headers") + if not headers: + headers = request_data.get("headers") + if not headers: + headers = (request_data.get("metadata") or {}).get("headers") + if not headers and logging_obj and getattr(logging_obj, "model_call_details", None): + headers = ( + (logging_obj.model_call_details or {}).get("litellm_params", {}).get("metadata", {}).get("headers") + ) + if not isinstance(headers, dict): + return None + + forwarded_header_names = ("shade_scan_id",) + forwarded_headers = {} + for key, value in headers.items(): + if str(key).lower() in forwarded_header_names: + forwarded_headers[str(key)] = str(value) + return forwarded_headers or None + def _prepare_payload( - self, messages: List[Dict[str, str]], dynamic_body: dict, request_data: dict - ) -> Optional[Dict[str, Any]]: - payload: Dict[str, Any] = {"messages": messages} + self, + messages: list[dict[str, str]], + dynamic_body: dict, + request_data: dict, + logging_obj: Optional["LiteLLMLoggingObj"] = None, + ) -> Optional[dict[str, Any]]: + payload: dict[str, Any] = {"messages": messages} categories = dynamic_body.get("categories") or self.categories if categories: @@ -523,10 +551,16 @@ class GraySwanGuardrail(CustomGuardrail): if "metadata" in dynamic_body: payload["metadata"] = dynamic_body["metadata"] + inbound_headers = self._extract_inbound_headers(request_data, logging_obj) + litellm_metadata = request_data.get("litellm_metadata") - if isinstance(litellm_metadata, dict) and litellm_metadata: - cleaned_litellm_metadata = dict(litellm_metadata) - # cleaned_litellm_metadata.pop("user_api_key_auth", None) + cleaned_litellm_metadata = dict(litellm_metadata) if isinstance(litellm_metadata, dict) else {} + if inbound_headers: + existing_headers = cleaned_litellm_metadata.get("headers") + cleaned_litellm_metadata["headers"] = ( + {**existing_headers, **inbound_headers} if isinstance(existing_headers, dict) else inbound_headers + ) + if cleaned_litellm_metadata: sanitized = safe_json_loads(safe_dumps(cleaned_litellm_metadata), default={}) if isinstance(sanitized, dict) and sanitized: payload["litellm_metadata"] = sanitized diff --git a/litellm/proxy/management_endpoints/key_management_endpoints.py b/litellm/proxy/management_endpoints/key_management_endpoints.py index 63f4b731871..bf64f537c7f 100644 --- a/litellm/proxy/management_endpoints/key_management_endpoints.py +++ b/litellm/proxy/management_endpoints/key_management_endpoints.py @@ -111,6 +111,7 @@ from litellm.repositories.verification_token_repository import ( VerificationTokenRepository, ) from litellm.router import Router +from litellm.secret_managers.base_secret_manager import raise_if_unsafe_secret_name from litellm.secret_managers.main import get_secret from litellm.types.proxy.management_endpoints.key_management_endpoints import ( BulkUpdateKeyRequest, @@ -6291,8 +6292,13 @@ def _validate_key_alias_format(key_alias: Optional[str]) -> None: """ Validate the format of the key_alias. - Gated behind ``litellm.enable_key_alias_format_validation`` (default **False**). - When disabled, no validation is performed so existing workflows are not broken. + A baseline validation always runs, regardless of + ``litellm.enable_key_alias_format_validation``. + + The remaining charset/length rules are gated behind + ``litellm.enable_key_alias_format_validation`` (default **False**). When disabled, + only the baseline validation above is performed, so existing workflows are not + broken. Rules (when enabled): - None is OK (no alias). @@ -6300,10 +6306,20 @@ def _validate_key_alias_format(key_alias: Optional[str]) -> None: - start/end with alphanumeric - only allow a-zA-Z0-9_-/.@ """ - if not litellm.enable_key_alias_format_validation: + if key_alias is None: return - if key_alias is None: + try: + raise_if_unsafe_secret_name(key_alias) + except ValueError: + raise ProxyException( + message="Invalid key_alias", + type=ProxyErrorTypes.bad_request_error, + param="key_alias", + code=400, + ) + + if not litellm.enable_key_alias_format_validation: return if not _KEY_ALIAS_PATTERN.match(key_alias): diff --git a/litellm/proxy/management_endpoints/mcp_management_endpoints.py b/litellm/proxy/management_endpoints/mcp_management_endpoints.py index c9952b245c7..907a17d76d9 100644 --- a/litellm/proxy/management_endpoints/mcp_management_endpoints.py +++ b/litellm/proxy/management_endpoints/mcp_management_endpoints.py @@ -125,7 +125,9 @@ if MCP_AVAILABLE: get_user_env_vars_bulk, get_user_oauth_credential, list_user_oauth_credentials, + mcp_oauth_token_identity, merge_user_env_vars, + purge_user_oauth_credentials_for_server, reject_mcp_server, store_user_credential, store_user_oauth_credential, @@ -2318,6 +2320,19 @@ if MCP_AVAILABLE: }, ) + # Snapshot the pre-update identity so we can detect a mint-relevant change below. The read is + # advisory (it only feeds the stale-token purge decision), so a failure skips the purge with a + # warning instead of failing the edit, whose primary job is the update itself. + try: + old_server_record = await get_mcp_server(prisma_client, payload.server_id) + except Exception as exc: # noqa: BLE001 - advisory read; invalidation is best-effort end-to-end + verbose_logger.warning( + "MCP server %s: could not snapshot the pre-update record; skipping the stale-token check: %s", + payload.server_id, + exc, + ) + old_server_record = None + # try to update the mcp server mcp_server_record_updated = await update_mcp_server( prisma_client, @@ -2336,6 +2351,30 @@ if MCP_AVAILABLE: # Ensure registry is up to date by reloading from database await global_mcp_server_manager.reload_servers_from_database() + # If a field that determines which upstream OAuth token gets minted changed (url/audience, OAuth + # mode/grant, authorization-server endpoints, or the OAuth client + scopes), every stored per-user + # token was minted for the old configuration and is stale. Purge them (DB + cache) so the next + # tool call re-authorizes instead of forwarding a token for a resource/AS/client that no longer + # matches. Best-effort: a purge failure must not fail the update, whose primary job already + # succeeded. + if old_server_record is not None and mcp_oauth_token_identity(old_server_record) != mcp_oauth_token_identity( + mcp_server_record_updated + ): + try: + purged = await purge_user_oauth_credentials_for_server(prisma_client, payload.server_id) + if purged: + verbose_logger.info( + "MCP server %s: purged %d stale per-user OAuth token(s) after a mint-relevant config change", + payload.server_id, + purged, + ) + except Exception as exc: # noqa: BLE001 - purge is best-effort; the server update already succeeded + verbose_logger.warning( + "MCP server %s: failed to purge stale per-user OAuth tokens after config change: %s", + payload.server_id, + exc, + ) + # TODO: Enterprise: Finish audit log trail if litellm.store_audit_logs: pass diff --git a/litellm/proxy/management_endpoints/ui_sso.py b/litellm/proxy/management_endpoints/ui_sso.py index dbf514d2298..065464aa565 100644 --- a/litellm/proxy/management_endpoints/ui_sso.py +++ b/litellm/proxy/management_endpoints/ui_sso.py @@ -113,7 +113,7 @@ from litellm.proxy.utils import ( from litellm.repositories.table_repositories import SSOConfigRepository from litellm.repositories.team_repository import TeamRepository from litellm.repositories.user_repository import UserRepository -from litellm.secret_managers.main import get_secret_bool, str_to_bool +from litellm.secret_managers.main import get_secret_bool, get_secret_str, str_to_bool from litellm.types.proxy.management_endpoints.ui_sso import * # noqa: F403 from litellm.types.proxy.management_endpoints.ui_sso import ( DefaultTeamSSOParams, @@ -1746,6 +1746,18 @@ async def auth_callback(request: Request, state: Optional[str] = None): """Verify login""" verbose_proxy_logger.info(f"Starting SSO callback with state: {state}") + oauth_error = request.query_params.get("error") + if oauth_error: + oauth_error_description = request.query_params.get("error_description") + verbose_proxy_logger.warning( + f"SSO callback received OAuth error: {oauth_error}, description: {oauth_error_description}" + ) + raise HTTPException( + status_code=401, + detail=f"OAuth error: {oauth_error}" + + (f", error_description: {oauth_error_description}" if oauth_error_description else ""), + ) + # Check if this is a CLI login (state starts with our CLI prefix) from litellm.constants import LITELLM_CLI_SESSION_TOKEN_PREFIX from litellm.proxy._types import LiteLLM_JWTAuth @@ -3737,8 +3749,7 @@ class MicrosoftSSOHandler: Handles Microsoft SSO callback response and returns a CustomOpenID object """ - graph_api_base_url = "https://graph.microsoft.com/v1.0" - graph_api_user_groups_endpoint = f"{graph_api_base_url}/me/memberOf" + DEFAULT_GRAPH_API_BASE_URL = "https://graph.microsoft.com/v1.0" """ Constants @@ -3748,6 +3759,19 @@ class MicrosoftSSOHandler: # used for debugging to show the user groups litellm found from Graph API GRAPH_API_RESPONSE_KEY = "graph_api_user_groups" + @staticmethod + def get_graph_api_base_url() -> str: + """ + Returns the Microsoft Graph API base URL, configurable via the + `MICROSOFT_GRAPH_ENDPOINT` env var so non-default clouds such as Azure + Government (GCC High) can point at `https://graph.microsoft.us/v1.0` + """ + return get_secret_str("MICROSOFT_GRAPH_ENDPOINT") or MicrosoftSSOHandler.DEFAULT_GRAPH_API_BASE_URL + + @staticmethod + def get_graph_api_user_groups_endpoint() -> str: + return f"{MicrosoftSSOHandler.get_graph_api_base_url()}/me/memberOf" + @staticmethod async def get_microsoft_callback_response( request: Request, @@ -3924,7 +3948,7 @@ class MicrosoftSSOHandler: # Fetch user membership from Microsoft Graph API all_group_ids = [] - next_link: Optional[str] = MicrosoftSSOHandler.graph_api_user_groups_endpoint + next_link: Optional[str] = MicrosoftSSOHandler.get_graph_api_user_groups_endpoint() auth_headers = {"Authorization": f"Bearer {access_token}"} page_count = 0 @@ -4007,7 +4031,7 @@ class MicrosoftSSOHandler: Users use Enterprise Applications to manage Groups and Users on Microsoft Entra ID """ - base_url = "https://graph.microsoft.com/v1.0" + base_url = MicrosoftSSOHandler.get_graph_api_base_url() # Endpoint to get app role assignments for the given service principal endpoint = f"/servicePrincipals/{service_principal_id}/appRoleAssignedTo" url = base_url + endpoint diff --git a/litellm/proxy/proxy_server.py b/litellm/proxy/proxy_server.py index 13e2d4f1252..4114bda47c9 100644 --- a/litellm/proxy/proxy_server.py +++ b/litellm/proxy/proxy_server.py @@ -8544,6 +8544,8 @@ async def chat_completion( except ModifyResponseException as e: # Guardrail flagged content in passthrough mode - return 200 with violation message _data = e.request_data + # Capture logging_obj before post_call_failure_hook pops it from _data. + _logging_obj = _data.get("litellm_logging_obj") await proxy_logging_obj.post_call_failure_hook( user_api_key_dict=user_api_key_dict, original_exception=e, @@ -8563,7 +8565,7 @@ async def chat_completion( completion_stream=_iterator, model=e.model, custom_llm_provider="cached_response", - logging_obj=_data.get("litellm_logging_obj", None), + logging_obj=_logging_obj, ) selected_data_generator = select_data_generator( response=_streaming_response, diff --git a/litellm/proxy/spend_tracking/spend_management_endpoints.py b/litellm/proxy/spend_tracking/spend_management_endpoints.py index 8f530e3b8ce..bce2e3581b5 100644 --- a/litellm/proxy/spend_tracking/spend_management_endpoints.py +++ b/litellm/proxy/spend_tracking/spend_management_endpoints.py @@ -1622,6 +1622,10 @@ async def ui_view_spend_logs( default=None, description="request_id to get spend logs for specific request_id", ), + session_id: str | None = fastapi.Query( + default=None, + description="Filter spend logs by session_id (partial string match)", + ), team_id: str | None = fastapi.Query( default=None, description="Filter spend logs by team_id", @@ -1906,6 +1910,12 @@ async def ui_view_spend_logs( p += 2 sql_conditions.append(or_clause) + if session_id is not None and isinstance(session_id, str): + like_escaped_session_id = session_id.replace("\\", "\\\\").replace("%", "\\%").replace("_", "\\_") + sql_conditions.append(f"session_id LIKE ${p}") + sql_params.append(f"%{like_escaped_session_id}%") + p += 1 + # Status filter if status_filter is not None: if status_filter == "success": diff --git a/litellm/realtime_api/main.py b/litellm/realtime_api/main.py index 0566ff73683..5ecf4d91ff6 100644 --- a/litellm/realtime_api/main.py +++ b/litellm/realtime_api/main.py @@ -515,6 +515,7 @@ async def _realtime_health_check( api_base: Optional[str] = None, api_version: Optional[str] = None, realtime_protocol: Optional[str] = None, + model_params: Optional[dict] = None, ): """ Health check for realtime API - tries connection to the realtime API websocket @@ -550,14 +551,17 @@ async def _realtime_health_check( elif custom_llm_provider == "xai": url = xai_realtime._construct_url(api_base=api_base or "https://api.x.ai/v1", query_params={"model": model}) elif custom_llm_provider == "vertex_ai": - vertex_location = litellm.vertex_location or get_secret_str("VERTEXAI_LOCATION") - resolved_location = vertex_llm_base.get_vertex_region(vertex_region=vertex_location, model=model) + vertex_model_params = model_params or {} + resolved_location = vertex_llm_base.get_vertex_region( + vertex_region=VertexBase.safe_get_vertex_ai_location(vertex_model_params), + model=model, + ) ( access_token, resolved_project, ) = await vertex_llm_base._ensure_access_token_async( - credentials=None, - project_id=litellm.vertex_project or get_secret_str("VERTEXAI_PROJECT"), + credentials=VertexBase.safe_get_vertex_ai_credentials(vertex_model_params), + project_id=VertexBase.safe_get_vertex_ai_project(vertex_model_params), custom_llm_provider="vertex_ai", ) vertex_realtime_config = VertexAIRealtimeConfig( diff --git a/litellm/rerank_api/main.py b/litellm/rerank_api/main.py index 9320c7fae8a..b6ebf5589b7 100644 --- a/litellm/rerank_api/main.py +++ b/litellm/rerank_api/main.py @@ -158,7 +158,7 @@ def rerank( instruction=instruction, non_default_params=kwargs, ) - verbose_logger.info(f"optional_rerank_params: {optional_rerank_params}") + verbose_logger.debug(f"optional_rerank_params: {optional_rerank_params}") if isinstance(optional_params.timeout, str): optional_params.timeout = float(optional_params.timeout) diff --git a/litellm/secret_managers/base_secret_manager.py b/litellm/secret_managers/base_secret_manager.py index d33d76093c9..2bb8dc73138 100644 --- a/litellm/secret_managers/base_secret_manager.py +++ b/litellm/secret_managers/base_secret_manager.py @@ -1,3 +1,4 @@ +import re from abc import ABC, abstractmethod from typing import Any, Dict, Optional, Union @@ -5,6 +6,20 @@ import httpx from litellm import verbose_logger +_UNSAFE_SECRET_NAME_PATTERN = re.compile(r"(^|/)\.\.(/|$)|[\x00-\x1f\x7f-\x9f…

]") + + +def raise_if_unsafe_secret_name(secret_name: str) -> None: + """ + Validate a secret name before it is used by a secret manager integration. + + Rejects ".." only as a path segment (bounded by "/" or the start/end of the + string, e.g. "../x", "x/..", or exactly ".."), not as a plain substring, so + names like "release-1.0..2" are not rejected. + """ + if _UNSAFE_SECRET_NAME_PATTERN.search(secret_name): + raise ValueError(f"Invalid secret_name {secret_name!r}") + class BaseSecretManager(ABC): """ diff --git a/litellm/secret_managers/cyberark_secret_manager.py b/litellm/secret_managers/cyberark_secret_manager.py index faf6224757f..2b888cb85f6 100644 --- a/litellm/secret_managers/cyberark_secret_manager.py +++ b/litellm/secret_managers/cyberark_secret_manager.py @@ -4,6 +4,7 @@ from typing import Any, Dict, Optional, Union from urllib.parse import quote import httpx +import yaml import litellm from litellm._logging import verbose_logger @@ -15,7 +16,7 @@ from litellm.llms.custom_httpx.http_handler import ( ) from litellm.proxy._types import KeyManagementSystem -from .base_secret_manager import BaseSecretManager +from .base_secret_manager import BaseSecretManager, raise_if_unsafe_secret_name from .main import str_to_bool @@ -125,8 +126,11 @@ class CyberArkSecretManager(BaseSecretManager): """ # In production, we'd check if the variable exists first # For now, we'll attempt to create it and ignore if it already exists + raise_if_unsafe_secret_name(secret_name) policy_url = f"{self.conjur_addr}/policies/{self.conjur_account}/policy/root" - policy_yaml = f"- !variable {secret_name}\n" + # Use a real YAML serializer to build the scalar safely. + quoted_name = yaml.safe_dump(secret_name, default_style='"').strip() + policy_yaml = f"- !variable {quoted_name}\n" try: client = _get_httpx_client(params={"ssl_verify": self.ssl_verify}) diff --git a/litellm/secret_managers/hashicorp_secret_manager.py b/litellm/secret_managers/hashicorp_secret_manager.py index bd1b1097347..039aecb9e58 100644 --- a/litellm/secret_managers/hashicorp_secret_manager.py +++ b/litellm/secret_managers/hashicorp_secret_manager.py @@ -14,7 +14,7 @@ from litellm.llms.custom_httpx.http_handler import ( ) from litellm.proxy._types import KeyManagementSystem -from .base_secret_manager import BaseSecretManager +from .base_secret_manager import BaseSecretManager, raise_if_unsafe_secret_name class HashicorpSecretManager(BaseSecretManager): @@ -220,6 +220,7 @@ class HashicorpSecretManager(BaseSecretManager): - With custom mount: http://127.0.0.1:8200/v1/kv/data/mykey - With path prefix: http://127.0.0.1:8200/v1/secret/data/myapp/mykey """ + raise_if_unsafe_secret_name(secret_name) resolved_namespace = self._sanitize_path_component(namespace if namespace is not None else self.vault_namespace) resolved_mount = self._sanitize_path_component(mount_name if mount_name is not None else self.vault_mount_name) if resolved_mount is None: diff --git a/litellm/types/mcp.py b/litellm/types/mcp.py index 9c564a3c7a6..d273e8ec4db 100644 --- a/litellm/types/mcp.py +++ b/litellm/types/mcp.py @@ -38,6 +38,8 @@ class MCPAuth(str, enum.Enum): aws_sigv4 = "aws_sigv4" token = "token" oauth2_token_exchange = "oauth2_token_exchange" + true_passthrough = "true_passthrough" + oauth_delegate = "oauth_delegate" # RFC 8693 default subject_token_type. A NULL column / omitted config key means @@ -60,6 +62,8 @@ MCPAuthType = Optional[ MCPAuth.aws_sigv4, MCPAuth.token, MCPAuth.oauth2_token_exchange, + MCPAuth.true_passthrough, + MCPAuth.oauth_delegate, ] ] diff --git a/litellm/types/mcp_server/mcp_server_manager.py b/litellm/types/mcp_server/mcp_server_manager.py index 522a6f09165..f102ab5b7b9 100644 --- a/litellm/types/mcp_server/mcp_server_manager.py +++ b/litellm/types/mcp_server/mcp_server_manager.py @@ -152,6 +152,18 @@ class MCPServer(BaseModel): """True if this is an OAuth2 server that relies on per-user tokens (no client_credentials).""" return self.auth_type == MCPAuth.oauth2 and not self.has_client_credentials + @property + def is_true_passthrough(self) -> bool: + """True for the transparent-proxy mode: LiteLLM performs no admission auth and forwards the + client's ``Authorization`` to the upstream unchanged.""" + return self.auth_type == MCPAuth.true_passthrough + + @property + def is_oauth_delegate(self) -> bool: + """True for the delegated-upstream-OAuth mode: LiteLLM still admits the caller (API key / SSO / + JWT) but forwards the caller's separate upstream ``Authorization`` unchanged, minting nothing.""" + return self.auth_type == MCPAuth.oauth_delegate + @property def requires_per_user_auth(self) -> bool: """ @@ -167,6 +179,9 @@ class MCPServer(BaseModel): if self.needs_user_oauth_token: return True + if self.is_true_passthrough or self.is_oauth_delegate: + return True + # PAT passthrough: auth_type is none but extra_headers includes auth headers if self.auth_type == MCPAuth.none and self.extra_headers: auth_header_names = {"authorization", "x-api-key", "api-key", "apikey"} diff --git a/litellm/types/utils.py b/litellm/types/utils.py index 2403ab8c4ec..63ed876e232 100644 --- a/litellm/types/utils.py +++ b/litellm/types/utils.py @@ -2523,6 +2523,23 @@ class StandardLoggingMCPToolCall(TypedDict, total=False): the client is driving a stateful session. Absent for stateless calls. """ + mcp_auth_mode: Optional[str] + """ + The server's auth_type for this call (e.g. `true_passthrough`, `oauth_delegate`, + `oauth2`). For the client-forwarded token modes this records that the caller's own + upstream token was relayed, so an audit can attribute a relayed request to its mode + without logging any credential. + """ + + mcp_server_resource: Optional[str] + """ + The origin (scheme + host + port) of the upstream MCP server the tool call was forwarded + to. Redacted for logging: userinfo, the path, the query string, and the fragment are all + stripped, because hosted MCP servers routinely embed the credential in the URL path and + this value is readable by callers via request logs. + Records which upstream received a relayed request; never a credential. + """ + class StandardLoggingVectorStoreRequest(TypedDict, total=False): """ @@ -3383,6 +3400,7 @@ class LlmProviders(str, Enum): LIBERTAI = "libertai" PINSTRIPES = "pinstripes" DARKBLOOM = "darkbloom" + META = "meta" LITELLM_AGENT = "litellm_agent" CURSOR = "cursor" BEDROCK_MANTLE = "bedrock_mantle" diff --git a/litellm/utils.py b/litellm/utils.py index 19c2fe16085..5af7b62b332 100644 --- a/litellm/utils.py +++ b/litellm/utils.py @@ -8028,6 +8028,16 @@ class ProviderConfigManager: ) return GithubCopilotAnthropicMessagesConfig() + + from litellm.llms.openai_like.json_loader import JSONProviderRegistry + + json_provider = JSONProviderRegistry.get(provider.value) + if json_provider is not None and "/v1/messages" in json_provider.supported_endpoints: + from litellm.llms.openai_like.messages.transformation import ( + JSONProviderAnthropicMessagesConfig, + ) + + return JSONProviderAnthropicMessagesConfig(json_provider) return None @staticmethod diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json index b961c326625..e79ddbe35d2 100644 --- a/model_prices_and_context_window.json +++ b/model_prices_and_context_window.json @@ -6012,6 +6012,522 @@ "supports_vision": true, "supports_web_search": true }, + "azure/gpt-5.6": { + "cache_read_input_token_cost": 5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1e-06, + "cache_read_input_token_cost_priority": 1e-06, + "cache_read_input_token_cost_above_272k_tokens_priority": 2e-06, + "input_cost_per_token": 5e-06, + "input_cost_per_token_above_272k_tokens": 1e-05, + "input_cost_per_token_priority": 1e-05, + "input_cost_per_token_above_272k_tokens_priority": 2e-05, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3e-05, + "output_cost_per_token_above_272k_tokens": 4.5e-05, + "output_cost_per_token_priority": 6e-05, + "output_cost_per_token_above_272k_tokens_priority": 9e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/gpt-5.6-sol": { + "cache_read_input_token_cost": 5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1e-06, + "cache_read_input_token_cost_priority": 1e-06, + "cache_read_input_token_cost_above_272k_tokens_priority": 2e-06, + "input_cost_per_token": 5e-06, + "input_cost_per_token_above_272k_tokens": 1e-05, + "input_cost_per_token_priority": 1e-05, + "input_cost_per_token_above_272k_tokens_priority": 2e-05, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3e-05, + "output_cost_per_token_above_272k_tokens": 4.5e-05, + "output_cost_per_token_priority": 6e-05, + "output_cost_per_token_above_272k_tokens_priority": 9e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/gpt-5.6-terra": { + "cache_read_input_token_cost": 2.5e-07, + "cache_read_input_token_cost_above_272k_tokens": 5e-07, + "cache_read_input_token_cost_priority": 5e-07, + "cache_read_input_token_cost_above_272k_tokens_priority": 1e-06, + "input_cost_per_token": 2.5e-06, + "input_cost_per_token_above_272k_tokens": 5e-06, + "input_cost_per_token_priority": 5e-06, + "input_cost_per_token_above_272k_tokens_priority": 1e-05, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 1.5e-05, + "output_cost_per_token_above_272k_tokens": 2.25e-05, + "output_cost_per_token_priority": 3e-05, + "output_cost_per_token_above_272k_tokens_priority": 4.5e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/gpt-5.6-luna": { + "cache_read_input_token_cost": 1e-07, + "cache_read_input_token_cost_above_272k_tokens": 2e-07, + "cache_read_input_token_cost_priority": 2e-07, + "cache_read_input_token_cost_above_272k_tokens_priority": 4e-07, + "input_cost_per_token": 1e-06, + "input_cost_per_token_above_272k_tokens": 2e-06, + "input_cost_per_token_priority": 2e-06, + "input_cost_per_token_above_272k_tokens_priority": 4e-06, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 6e-06, + "output_cost_per_token_above_272k_tokens": 9e-06, + "output_cost_per_token_priority": 1.2e-05, + "output_cost_per_token_above_272k_tokens_priority": 1.8e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/us/gpt-5.6": { + "cache_read_input_token_cost": 5.5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1.1e-06, + "cache_read_input_token_cost_priority": 1.375e-06, + "input_cost_per_token": 5.5e-06, + "input_cost_per_token_above_272k_tokens": 1.1e-05, + "input_cost_per_token_priority": 1.375e-05, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3.3e-05, + "output_cost_per_token_above_272k_tokens": 4.95e-05, + "output_cost_per_token_priority": 8.25e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/us/gpt-5.6-sol": { + "cache_read_input_token_cost": 5.5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1.1e-06, + "cache_read_input_token_cost_priority": 1.375e-06, + "input_cost_per_token": 5.5e-06, + "input_cost_per_token_above_272k_tokens": 1.1e-05, + "input_cost_per_token_priority": 1.375e-05, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3.3e-05, + "output_cost_per_token_above_272k_tokens": 4.95e-05, + "output_cost_per_token_priority": 8.25e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/us/gpt-5.6-terra": { + "cache_read_input_token_cost": 2.75e-07, + "cache_read_input_token_cost_above_272k_tokens": 5.5e-07, + "cache_read_input_token_cost_priority": 6.875e-07, + "input_cost_per_token": 2.75e-06, + "input_cost_per_token_above_272k_tokens": 5.5e-06, + "input_cost_per_token_priority": 6.875e-06, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 1.65e-05, + "output_cost_per_token_above_272k_tokens": 2.475e-05, + "output_cost_per_token_priority": 4.125e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/us/gpt-5.6-luna": { + "cache_read_input_token_cost": 1.1e-07, + "cache_read_input_token_cost_above_272k_tokens": 2.2e-07, + "cache_read_input_token_cost_priority": 2.75e-07, + "input_cost_per_token": 1.1e-06, + "input_cost_per_token_above_272k_tokens": 2.2e-06, + "input_cost_per_token_priority": 2.75e-06, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 6.6e-06, + "output_cost_per_token_above_272k_tokens": 9.9e-06, + "output_cost_per_token_priority": 1.65e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/eu/gpt-5.6": { + "cache_read_input_token_cost": 5.5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1.1e-06, + "cache_read_input_token_cost_priority": 1.375e-06, + "input_cost_per_token": 5.5e-06, + "input_cost_per_token_above_272k_tokens": 1.1e-05, + "input_cost_per_token_priority": 1.375e-05, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3.3e-05, + "output_cost_per_token_above_272k_tokens": 4.95e-05, + "output_cost_per_token_priority": 8.25e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/eu/gpt-5.6-sol": { + "cache_read_input_token_cost": 5.5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1.1e-06, + "cache_read_input_token_cost_priority": 1.375e-06, + "input_cost_per_token": 5.5e-06, + "input_cost_per_token_above_272k_tokens": 1.1e-05, + "input_cost_per_token_priority": 1.375e-05, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3.3e-05, + "output_cost_per_token_above_272k_tokens": 4.95e-05, + "output_cost_per_token_priority": 8.25e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/eu/gpt-5.6-terra": { + "cache_read_input_token_cost": 2.75e-07, + "cache_read_input_token_cost_above_272k_tokens": 5.5e-07, + "cache_read_input_token_cost_priority": 6.875e-07, + "input_cost_per_token": 2.75e-06, + "input_cost_per_token_above_272k_tokens": 5.5e-06, + "input_cost_per_token_priority": 6.875e-06, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 1.65e-05, + "output_cost_per_token_above_272k_tokens": 2.475e-05, + "output_cost_per_token_priority": 4.125e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, + "azure/eu/gpt-5.6-luna": { + "cache_read_input_token_cost": 1.1e-07, + "cache_read_input_token_cost_above_272k_tokens": 2.2e-07, + "cache_read_input_token_cost_priority": 2.75e-07, + "input_cost_per_token": 1.1e-06, + "input_cost_per_token_above_272k_tokens": 2.2e-06, + "input_cost_per_token_priority": 2.75e-06, + "litellm_provider": "azure", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 6.6e-06, + "output_cost_per_token_above_272k_tokens": 9.9e-06, + "output_cost_per_token_priority": 1.65e-05, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_none_reasoning_effort": true, + "supports_xhigh_reasoning_effort": true, + "supports_minimal_reasoning_effort": false + }, "azure/gpt-5.5": { "cache_read_input_token_cost": 5e-07, "cache_read_input_token_cost_above_272k_tokens": 1e-06, @@ -22431,6 +22947,218 @@ "supports_xhigh_reasoning_effort": true, "supports_minimal_reasoning_effort": true }, + "gpt-5.6": { + "cache_creation_input_token_cost": 6.25e-06, + "cache_creation_input_token_cost_above_272k_tokens": 1.25e-05, + "cache_creation_input_token_cost_flex": 3.125e-06, + "cache_creation_input_token_cost_priority": 1.25e-05, + "cache_read_input_token_cost": 5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1e-06, + "cache_read_input_token_cost_flex": 2.5e-07, + "cache_read_input_token_cost_priority": 1e-06, + "input_cost_per_token": 5e-06, + "input_cost_per_token_above_272k_tokens": 1e-05, + "input_cost_per_token_batches": 2.5e-06, + "input_cost_per_token_flex": 2.5e-06, + "input_cost_per_token_priority": 1e-05, + "litellm_provider": "openai", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3e-05, + "output_cost_per_token_above_272k_tokens": 4.5e-05, + "output_cost_per_token_batches": 1.5e-05, + "output_cost_per_token_flex": 1.5e-05, + "output_cost_per_token_priority": 6e-05, + "regional_processing_uplift_multiplier_eu": 1.1, + "regional_processing_uplift_multiplier_us": 1.1, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_minimal_reasoning_effort": false, + "supports_native_streaming": true, + "supports_none_reasoning_effort": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_xhigh_reasoning_effort": true + }, + "gpt-5.6-sol": { + "cache_creation_input_token_cost": 6.25e-06, + "cache_creation_input_token_cost_above_272k_tokens": 1.25e-05, + "cache_creation_input_token_cost_flex": 3.125e-06, + "cache_creation_input_token_cost_priority": 1.25e-05, + "cache_read_input_token_cost": 5e-07, + "cache_read_input_token_cost_above_272k_tokens": 1e-06, + "cache_read_input_token_cost_flex": 2.5e-07, + "cache_read_input_token_cost_priority": 1e-06, + "input_cost_per_token": 5e-06, + "input_cost_per_token_above_272k_tokens": 1e-05, + "input_cost_per_token_batches": 2.5e-06, + "input_cost_per_token_flex": 2.5e-06, + "input_cost_per_token_priority": 1e-05, + "litellm_provider": "openai", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 3e-05, + "output_cost_per_token_above_272k_tokens": 4.5e-05, + "output_cost_per_token_batches": 1.5e-05, + "output_cost_per_token_flex": 1.5e-05, + "output_cost_per_token_priority": 6e-05, + "regional_processing_uplift_multiplier_eu": 1.1, + "regional_processing_uplift_multiplier_us": 1.1, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_minimal_reasoning_effort": false, + "supports_native_streaming": true, + "supports_none_reasoning_effort": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_xhigh_reasoning_effort": true + }, + "gpt-5.6-terra": { + "cache_creation_input_token_cost": 3.125e-06, + "cache_creation_input_token_cost_above_272k_tokens": 6.25e-06, + "cache_creation_input_token_cost_flex": 1.5625e-06, + "cache_creation_input_token_cost_priority": 6.25e-06, + "cache_read_input_token_cost": 2.5e-07, + "cache_read_input_token_cost_above_272k_tokens": 5e-07, + "cache_read_input_token_cost_flex": 1.25e-07, + "cache_read_input_token_cost_priority": 5e-07, + "input_cost_per_token": 2.5e-06, + "input_cost_per_token_above_272k_tokens": 5e-06, + "input_cost_per_token_batches": 1.25e-06, + "input_cost_per_token_flex": 1.25e-06, + "input_cost_per_token_priority": 5e-06, + "litellm_provider": "openai", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 1.5e-05, + "output_cost_per_token_above_272k_tokens": 2.25e-05, + "output_cost_per_token_batches": 7.5e-06, + "output_cost_per_token_flex": 7.5e-06, + "output_cost_per_token_priority": 3e-05, + "regional_processing_uplift_multiplier_eu": 1.1, + "regional_processing_uplift_multiplier_us": 1.1, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_minimal_reasoning_effort": false, + "supports_native_streaming": true, + "supports_none_reasoning_effort": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_xhigh_reasoning_effort": true + }, + "gpt-5.6-luna": { + "cache_creation_input_token_cost": 1.25e-06, + "cache_creation_input_token_cost_above_272k_tokens": 2.5e-06, + "cache_creation_input_token_cost_flex": 6.25e-07, + "cache_creation_input_token_cost_priority": 2.5e-06, + "cache_read_input_token_cost": 1e-07, + "cache_read_input_token_cost_above_272k_tokens": 2e-07, + "cache_read_input_token_cost_flex": 5e-08, + "cache_read_input_token_cost_priority": 2e-07, + "input_cost_per_token": 1e-06, + "input_cost_per_token_above_272k_tokens": 2e-06, + "input_cost_per_token_batches": 5e-07, + "input_cost_per_token_flex": 5e-07, + "input_cost_per_token_priority": 2e-06, + "litellm_provider": "openai", + "max_input_tokens": 1050000, + "max_output_tokens": 128000, + "max_tokens": 128000, + "mode": "chat", + "output_cost_per_token": 6e-06, + "output_cost_per_token_above_272k_tokens": 9e-06, + "output_cost_per_token_batches": 3e-06, + "output_cost_per_token_flex": 3e-06, + "output_cost_per_token_priority": 1.2e-05, + "regional_processing_uplift_multiplier_eu": 1.1, + "regional_processing_uplift_multiplier_us": 1.1, + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/batch", + "/v1/responses" + ], + "supported_modalities": [ + "text", + "image" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_minimal_reasoning_effort": false, + "supports_native_streaming": true, + "supports_none_reasoning_effort": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_xhigh_reasoning_effort": true + }, "gpt-5.5": { "cache_read_input_token_cost": 5e-07, "cache_read_input_token_cost_above_272k_tokens": 1e-06, @@ -25447,6 +26175,42 @@ "supports_function_calling": true, "supports_tool_choice": false }, + "meta/muse-spark-1.1": { + "cache_read_input_token_cost": 1.5e-07, + "input_cost_per_token": 1.25e-06, + "litellm_provider": "meta", + "max_input_tokens": 1048576, + "max_output_tokens": 131072, + "max_tokens": 131072, + "mode": "chat", + "output_cost_per_token": 4.25e-06, + "source": "https://dev.meta.ai/docs/getting-started/pricing-rate-limits", + "supported_endpoints": [ + "/v1/chat/completions", + "/v1/responses", + "/v1/messages" + ], + "supported_modalities": [ + "text", + "image", + "video" + ], + "supported_output_modalities": [ + "text" + ], + "supports_function_calling": true, + "supports_minimal_reasoning_effort": true, + "supports_parallel_function_calling": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_system_messages": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true, + "supports_xhigh_reasoning_effort": true + }, "meta_llama/Llama-3.3-70B-Instruct": { "litellm_provider": "meta_llama", "max_input_tokens": 128000, @@ -38303,6 +39067,48 @@ "supports_vision": true, "supports_web_search": true }, + "xai/grok-4.5": { + "cache_read_input_token_cost": 5e-07, + "cache_read_input_token_cost_above_200k_tokens": 1e-06, + "input_cost_per_token": 2e-06, + "input_cost_per_token_above_200k_tokens": 4e-06, + "litellm_provider": "xai", + "max_input_tokens": 500000, + "max_output_tokens": 500000, + "max_tokens": 500000, + "mode": "chat", + "output_cost_per_token": 6e-06, + "output_cost_per_token_above_200k_tokens": 1.2e-05, + "source": "https://docs.x.ai/docs/models", + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true + }, + "xai/grok-4.5-latest": { + "cache_read_input_token_cost": 5e-07, + "cache_read_input_token_cost_above_200k_tokens": 1e-06, + "input_cost_per_token": 2e-06, + "input_cost_per_token_above_200k_tokens": 4e-06, + "litellm_provider": "xai", + "max_input_tokens": 500000, + "max_output_tokens": 500000, + "max_tokens": 500000, + "mode": "chat", + "output_cost_per_token": 6e-06, + "output_cost_per_token_above_200k_tokens": 1.2e-05, + "source": "https://docs.x.ai/docs/models", + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_vision": true, + "supports_web_search": true + }, "xai/grok-beta": { "input_cost_per_token": 5e-06, "litellm_provider": "xai", diff --git a/provider_endpoints_support.json b/provider_endpoints_support.json index 3034ada56ba..65db63dc045 100644 --- a/provider_endpoints_support.json +++ b/provider_endpoints_support.json @@ -1984,6 +1984,23 @@ "interactions": true } }, + "meta": { + "display_name": "Meta Model API (`meta`)", + "url": "https://docs.litellm.ai/docs/providers/meta", + "endpoints": { + "chat_completions": true, + "messages": true, + "responses": true, + "embeddings": false, + "image_generations": false, + "audio_transcriptions": false, + "audio_speech": false, + "moderations": false, + "batches": false, + "rerank": false, + "a2a": false + } + }, "pinstripes": { "display_name": "Pinstripes (`pinstripes`)", "url": "https://docs.litellm.ai/docs/providers/pinstripes", diff --git a/pyproject.toml b/pyproject.toml index 3f5458c5494..4c70bf66031 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -63,7 +63,7 @@ proxy = [ "azure-storage-blob>=12.28.0,<13.0", "mcp>=1.26.0,<2.0", "litellm-proxy-extras==0.4.75", - "litellm-enterprise==0.1.48", + "litellm-enterprise==0.1.49", "RestrictedPython>=8.1,<9.0", "rich>=13.9.4,<14.0", "polars>=1.38.1,<2.0", @@ -259,6 +259,7 @@ constraint-dependencies = [ "tornado>=6.5.6", "aiohttp>=3.14.1,<4.0", "packaging>=24.0", + "soupsieve>=2.8.4", ] override-dependencies = [ # a2a-sdk 1.x requires packaging>=24.0; lunary 1.4.x still caps at <24.0. diff --git a/tests/benchmarks/conftest.py b/tests/benchmarks/conftest.py new file mode 100644 index 00000000000..c9b31cfb7d7 --- /dev/null +++ b/tests/benchmarks/conftest.py @@ -0,0 +1,36 @@ +"""Shared setup keeping CodSpeed measurements hermetic. + +CodSpeed's callgrind instrumentation counts instructions from every thread while +a measurement window is open, and valgrind serializes all threads onto one +virtual CPU. Work deferred to litellm's shared logging executor would therefore +be attributed to whichever benchmark the valgrind scheduler resumes it under, +flipping results between runs. Running the executor inline keeps each +benchmark's cost self-contained and deterministic. +""" + +from collections.abc import Callable, Iterator +from concurrent.futures import Future +from typing import ParamSpec, TypeVar + +import pytest + +from litellm.litellm_core_utils.thread_pool_executor import executor + +P = ParamSpec("P") +R = TypeVar("R") + + +def _submit_inline(fn: Callable[P, R], /, *args: P.args, **kwargs: P.kwargs) -> Future[R]: + future: Future[R] = Future() + try: + future.set_result(fn(*args, **kwargs)) + except BaseException as exc: + future.set_exception(exc) + return future + + +@pytest.fixture(autouse=True, scope="session") +def inline_logging_executor() -> Iterator[None]: + executor.submit = _submit_inline + yield + del executor.submit diff --git a/tests/benchmarks/test_benchmarks.py b/tests/benchmarks/test_benchmarks.py index 123dad93e11..59b3e0b6d5c 100644 --- a/tests/benchmarks/test_benchmarks.py +++ b/tests/benchmarks/test_benchmarks.py @@ -6,10 +6,13 @@ in the litellm hot path: token counting, model info lookup, provider resolution, and cost calculation. """ +import threading + import pytest import litellm from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider +from litellm.litellm_core_utils.thread_pool_executor import executor from litellm.litellm_core_utils.token_counter import token_counter @@ -205,3 +208,21 @@ def test_get_model_cost_key_exact_match(): def test_get_model_cost_key_case_insensitive(): """Benchmark model cost key lookup with case-insensitive fallback.""" litellm.utils._get_model_cost_key("GPT-4o") + + +# --------------------------------------------------------------------------- +# Measurement hermeticity guard +# --------------------------------------------------------------------------- + + +@pytest.mark.benchmark +def test_logging_executor_runs_inline(): + """Guard that the shared logging executor runs submissions inline. + + Deferred submissions execute on worker threads, and callgrind attributes + their instructions to whichever benchmark's measurement window is open when + the valgrind scheduler resumes them, making results nondeterministic. + """ + future = executor.submit(threading.get_ident) + assert future.done() + assert future.result() == threading.get_ident() diff --git a/tests/code_coverage_tests/recursive_detector.py b/tests/code_coverage_tests/recursive_detector.py index 6d0e12314f0..2af14e1e544 100644 --- a/tests/code_coverage_tests/recursive_detector.py +++ b/tests/code_coverage_tests/recursive_detector.py @@ -36,6 +36,7 @@ IGNORE_FUNCTIONS = [ "_collect_argument_paths", # max depth set. "_split_text", # max depth set. "_mask_sequence", # max depth set. + "_walk_payload", # max depth set (DEFAULT_MAX_RECURSE_DEPTH_SENSITIVE_DATA_MASKER). "_delete_nested_value_custom", # max depth set (bounded by number of path segments). "filter_exceptions_from_params", # max depth set (default 20) to prevent infinite recursion. "__getattr__", # lazy loading pattern in litellm/__init__.py with proper caching to prevent infinite recursion. diff --git a/tests/e2e/CLAUDE.md b/tests/e2e/CLAUDE.md index a4c507ca5ea..502c881c764 100644 --- a/tests/e2e/CLAUDE.md +++ b/tests/e2e/CLAUDE.md @@ -63,7 +63,7 @@ The harness is fully typed and new code must not add `Any` or widen the basedpyr The set of tests we want is a registry checked into this repo, one row per behavior; that file is the definition of done and the denominator. Each e2e test declares what it covers with `@pytest.mark.covers("...")`, and a small collector diffs the registry against the tests and ships coverage to the existing Grafana. No Allure, no new dependencies -Coverage is organized as module > feature > test. Dashboard modules are Core LLMs, Non-Core LLMs, MCPs, Management/UI, Reliability & Performance, Logging & Guardrails, and Other. A feature is either an endpoint (`/chat/completions`) or a behavior (fallbacks, rate limits; config-driven, with no route of its own). A cell reads like `llm.chat_completions.bedrock_converse.tool_use.stream.works` +Coverage is organized as module > feature > test. Dashboard modules are `Core LLMs`, `Non-Core LLMs`, `MCPs`, `Management/UI`, `Reliability & Performance`, `Logging & Guardrails`, and `Other`. The Loki stdout formatter maps those display modules to log-safe labels (`core_llms`, `non_core_llms`, `mcp`, `management_ui`, `reliability_performance`, `logging_guardrails`, and `other`) without changing JSON or Prometheus labels. A feature is either an endpoint (`/chat/completions`) or a behavior (fallbacks, rate limits; config-driven, with no route of its own). A cell reads like `llm.chat_completions.bedrock_converse.tool_use.stream.works` The metric is coverage: the share of registry rows that have a passing covering test, reported to Grafana per module so a gap surfaces as an uncovered row rather than a silent absence @@ -71,7 +71,7 @@ Tests do not declare a dashboard module directly. They only declare the registry ### Naming grammar per module -LLMs - endpoint features (subject = the route), seeded from the Claude Code compat matrix. `chat_completions`, `messages`, and `responses` are Core LLMs. Other LLM endpoints, including `batches` and `realtime`, roll up as Non-Core LLMs. +LLMs - endpoint features (subject = the route), seeded from the Claude Code compat matrix. `chat_completions`, `messages`, and `responses` roll up to `Core LLMs`. Other LLM endpoints, including `batches` and `realtime`, roll up to `Non-Core LLMs`. ``` llm..... diff --git a/tests/e2e/batches/capabilities.py b/tests/e2e/batches/capabilities.py index 522e3162e24..51e69fb5805 100644 --- a/tests/e2e/batches/capabilities.py +++ b/tests/e2e/batches/capabilities.py @@ -44,10 +44,10 @@ class Provider: ) case "azure": return LiteLLMParamsBody( - model="azure/gpt-4.1-mini-batch", + model="azure/gpt-5.4-mini-batch", api_base="os.environ/AZURE_API_BASE", api_key="os.environ/AZURE_API_KEY", - api_version="2024-07-01-preview", + api_version="2025-04-01-preview", ) case "vertex_ai": return LiteLLMParamsBody( @@ -97,7 +97,7 @@ class Capability: PROVIDERS: tuple[Provider, ...] = ( Provider("openai", "openai-batch", "gpt-4o-mini", can_cancel=True, can_list=True), - Provider("azure", "azure-batch", "gpt-4.1-mini-batch", can_cancel=True, can_list=True), + Provider("azure", "azure-batch", "gpt-5.4-mini-batch", can_cancel=True, can_list=True), Provider( "vertex_ai", "vertex-batch", "gemini-2.5-flash", can_cancel=True, can_list=True ), diff --git a/tests/e2e/bob_the_builder.py b/tests/e2e/bob_the_builder.py new file mode 100644 index 00000000000..18aff2edc98 --- /dev/null +++ b/tests/e2e/bob_the_builder.py @@ -0,0 +1,247 @@ +"""Bob the builder: on a red e2e run, ask Devin to fix the failing tests. + +Wired as a ``pytest_sessionfinish`` step (see ``conftest.py``). When the run went +red and remediation is enabled, it hands the failing tests plus their captured +tracebacks to Devin *through the LiteLLM proxy's own MCP gateway* -- the same +gateway + master key the suite already uses -- so Devin files a Linear ticket per +failure and opens fix PRs. Nothing new ships in the runner pod: the proxy already +registers the ``devin`` MCP server and holds ``DEVIN_API_KEY``, injecting it +upstream, so this process only needs the proxy key it always has. + +Opt-in via ``E2E_DEVIN_REMEDIATION=1`` so a normal local ``pytest tests/e2e`` run +never spawns a Devin session. ``DEVIN_DRY_RUN=1`` prints the prompt it would send +and makes no call. Everything is best-effort: any error here is logged and +swallowed so the run's exit status still reflects the tests, not remediation. +""" + +from __future__ import annotations + +import hashlib +import os +from collections.abc import Mapping, Sequence +from dataclasses import dataclass +from pathlib import Path +from typing import Protocol, cast + +import pytest +from pydantic import BaseModel, ConfigDict + +from e2e_config import MASTER_KEY, PROXY_BASE_URL +from e2e_http import Success +from transport import HttpTransport + +REMEDIATION_ENV = "E2E_DEVIN_REMEDIATION" +_LIST_PATH = "/mcp-rest/tools/list" +_CALL_PATH = "/mcp-rest/tools/call" + + +@dataclass(frozen=True, slots=True) +class Failure: + """One failed test: its pytest node id and the captured failure text.""" + + nodeid: str + detail: str + + +@dataclass(frozen=True, slots=True) +class Config: + server: str + create_tool: str + linear_team: str + target_repo: str + target_ref: str + max_failures: int + max_detail_chars: int + tags: tuple[str, ...] + dry_run: bool + + +class _NoParams(BaseModel): + pass + + +class _McpToolInfo(BaseModel): + model_config = ConfigDict(extra="allow") + server_name: str | None = None + alias: str | None = None + + +class _McpTool(BaseModel): + model_config = ConfigDict(extra="allow") + name: str + mcp_info: _McpToolInfo | None = None + + +class _McpToolsList(BaseModel): + model_config = ConfigDict(extra="allow") + tools: tuple[_McpTool, ...] = () + + +class _DevinSessionArgs(BaseModel): + prompt: str + title: str + tags: list[str] + + +class _ToolCallBody(BaseModel): + name: str + arguments: _DevinSessionArgs + + +class _ToolCallResult(BaseModel): + model_config = ConfigDict(extra="allow") + + +class _Report(Protocol): + @property + def nodeid(self) -> str: ... + + @property + def longreprtext(self) -> str: ... + + +class _TerminalReporter(Protocol): + stats: Mapping[str, Sequence[_Report]] + + +def _env(name: str, default: str) -> str: + value = os.environ.get(name, "").strip() + return value or default + + +def load_config() -> Config: + raw_tags = _env("DEVIN_TAGS", "e2e,stage") + return Config( + server=_env("DEVIN_MCP_SERVER", "devin"), + create_tool=_env("DEVIN_SESSION_TOOL", "devin_session_create"), + linear_team=_env("DEVIN_LINEAR_TEAM", "LIT"), + target_repo=_env("DEVIN_TARGET_REPO", "BerriAI/litellm"), + target_ref=_env("DEVIN_TARGET_REF", "litellm_internal_staging"), + max_failures=int(_env("DEVIN_MAX_FAILURES", "50")), + max_detail_chars=int(_env("DEVIN_MAX_DETAIL_CHARS", "3000")), + tags=tuple(t.strip() for t in raw_tags.split(",") if t.strip()), + dry_run=_env("DEVIN_DRY_RUN", "0") == "1", + ) + + +def collect_failures(session: pytest.Session, max_detail_chars: int) -> tuple[Failure, ...]: + """Pull the failed and errored tests (with their tracebacks) off the run's + terminal reporter. Returns empty when nothing failed or the reporter is + absent (e.g. a skipped, proxy-less session).""" + plugin: object = session.config.pluginmanager.getplugin("terminalreporter") + if plugin is None: + return () + reporter = cast(_TerminalReporter, plugin) + reports = (*reporter.stats.get("failed", ()), *reporter.stats.get("error", ())) + return tuple( + Failure(nodeid=r.nodeid, detail=r.longreprtext.strip()[-max_detail_chars:]) for r in reports + ) + + +def dedup_tag(failures: tuple[Failure, ...]) -> str: + """Stable short tag identifying this exact set of failing tests, so repeated + nightly runs on the same failures reference one body of work.""" + joined = "\n".join(sorted(f.nodeid for f in failures)) + return "e2e-fail-" + hashlib.sha256(joined.encode()).hexdigest()[:12] + + +def _revision() -> str: + for candidate in (Path(__file__).parent / ".litellm-revision", Path("/app/e2e/.litellm-revision")): + try: + return candidate.read_text(encoding="utf-8").strip() + except OSError: + continue + return _env("E2E_REVISION", "unknown") + + +def build_prompt(cfg: Config, failures: tuple[Failure, ...], tag: str) -> str: + shown = failures[: cfg.max_failures] + header = ( + f"The LiteLLM end-to-end suite failed on the " + f"{_env('E2E_ENVIRONMENT', 'stage')} proxy. Source repo {cfg.target_repo} " + f"at revision {_revision()} (branch {cfg.target_ref}). {len(failures)} " + f"test(s) failed" + + (f"; the first {len(shown)} are shown" if len(shown) < len(failures) else "") + + ".\n\n" + ) + task = ( + "For each failing test below:\n" + f"1. Open a Linear ticket under the {cfg.linear_team} team describing the " + "failure (test id, the assertion/error, likely cause), unless an open " + "ticket for that same test already exists -- do not create duplicates.\n" + f"2. Fix it in {cfg.target_repo}, branching off {cfg.target_ref} and " + "following the repo's CONTRIBUTING and CLAUDE.md conventions (meaningful " + "regression coverage, conventional commits, run the suite locally), then " + "open a PR that references the Linear ticket.\n" + "3. Prefer one focused PR per failing test; if several share a root cause, " + "group them and say so.\n" + f"Before starting, search existing sessions/PRs tagged '{tag}' or " + "referencing these test ids and continue that work instead of restarting.\n\n" + "Failing tests and their captured output:\n" + ) + blocks = [f"### {i}. {f.nodeid}\n```\n{f.detail}\n```\n" for i, f in enumerate(shown, start=1)] + return header + task + "\n".join(blocks) + + +def _resolve_tool_name(transport: HttpTransport, cfg: Config) -> str | None: + """Find Devin's create-session tool on the gateway. The proxy prefixes tools + with the server alias, so match by suffix and (when present) the owning + server.""" + result = transport.get( + _LIST_PATH, headers=transport.master, params=_NoParams(), response_type=_McpToolsList + ) + if not isinstance(result, Success): + print(f"bob_the_builder: could not list gateway MCP tools: {result}") + return None + for tool in result.data.tools: + owner = tool.mcp_info.server_name or tool.mcp_info.alias if tool.mcp_info else None + if (owner is None or owner == cfg.server) and ( + tool.name == cfg.create_tool or tool.name.endswith(cfg.create_tool) + ): + return tool.name + print( + f"bob_the_builder: no '{cfg.create_tool}' tool for server '{cfg.server}' on the gateway; " + f"saw {[t.name for t in result.data.tools]}" + ) + return None + + +def remediate(session: pytest.Session) -> None: + """Entry point called from ``pytest_sessionfinish``. No-op unless remediation + is enabled and the run actually had failures.""" + if os.environ.get(REMEDIATION_ENV) != "1": + return + cfg = load_config() + failures = collect_failures(session, cfg.max_detail_chars) + if not failures: + return + + tag = dedup_tag(failures) + title = f"Fix {len(failures)} failing LiteLLM e2e test(s) [{tag}]" + prompt = build_prompt(cfg, failures, tag) + args = _DevinSessionArgs(prompt=prompt, title=title, tags=[*cfg.tags, tag]) + + if cfg.dry_run: + print("bob_the_builder: DRY RUN -- would create a Devin session:") + print(f" server : {cfg.server}\n tool : {cfg.create_tool}\n title : {title}") + print(f" tags : {args.tags}\n---- prompt ----\n{prompt}") + return + + try: + transport = HttpTransport(base_url=PROXY_BASE_URL, master_key=MASTER_KEY) + tool_name = _resolve_tool_name(transport, cfg) + if tool_name is None: + return + result = transport.post( + _CALL_PATH, + headers=transport.master, + json=_ToolCallBody(name=tool_name, arguments=args), + response_type=_ToolCallResult, + ) + if isinstance(result, Success): + print(f"bob_the_builder: created Devin session for {len(failures)} failure(s) [{tag}]") + print(result.data.model_dump_json()) + else: + print(f"bob_the_builder: Devin session call failed: {result}") + except Exception as exc: # noqa: BLE001 - remediation must never fail the run + print(f"bob_the_builder: remediation error (ignored): {exc}") diff --git a/tests/e2e/conftest.py b/tests/e2e/conftest.py index 9ca5840df24..82f2604d492 100644 --- a/tests/e2e/conftest.py +++ b/tests/e2e/conftest.py @@ -107,6 +107,13 @@ def pytest_sessionfinish(session: pytest.Session, exitstatus: int) -> None: if spend_dir in sys.path: sys.path.remove(spend_dir) + try: + from bob_the_builder import remediate + + remediate(session) + except Exception as exc: # noqa: BLE001 - remediation is best-effort + print(f"devin remediation skipped: {exc}") + @pytest.fixture def resources(client: GatewayProvider) -> Iterator[ResourceManager]: diff --git a/tests/e2e/coverage_registry/README.md b/tests/e2e/coverage_registry/README.md index 4177cba7766..863ce34694d 100644 --- a/tests/e2e/coverage_registry/README.md +++ b/tests/e2e/coverage_registry/README.md @@ -9,18 +9,18 @@ note; the naming grammar lives in `tests/e2e/CLAUDE.md`. A **cell** is one customer-noticeable behavior a single e2e test can assert pass/fail on, for example `llm.chat_completions.bedrock_converse.tool_use.stream.works`. Cells are -grouped `module > feature > test`, with LLM cells split into Core LLMs and Non-Core -LLMs for dashboarding. Each cell carries a tier (P0/P1/P2), a source, and a +grouped `module > feature > test`, with LLM cells split into `Core LLMs` and +`Non-Core LLMs` for dashboarding. Each cell carries a tier (P0/P1/P2), a source, and a `fail_before_fix` flag. The rows live in per-prefix YAML files (`llm_*.yaml`, `mgmt.yaml`, `mcp.yaml`, `reliability.yaml`, `logging.yaml`, `guardrail.yaml`, `other.yaml`) and validate against the discriminated union in `schema.py`, so an LLM row cannot carry a guardrail field and vice versa. `llm` rows with `subject_endpoint` of `chat_completions`, `messages`, or -`responses` roll up to "Core LLMs"; all other LLM endpoints roll up to "Non-Core -LLMs". LLM endpoint, route, and capability values are typed in `schema.py`, so new -taxonomy values require an explicit schema change. `logging` and `guardrail` are two -id-prefixes that roll up into the single "Logging & Guardrails" dashboard module. +`responses` roll up to `Core LLMs`; all other LLM endpoints roll up to `Non-Core LLMs`. +LLM endpoint, route, and capability values are typed in `schema.py`, so new taxonomy +values require an explicit schema change. `logging` and `guardrail` are two id-prefixes +that roll up into the single `Logging & Guardrails` dashboard module. A test declares what it covers with a marker: @@ -40,8 +40,17 @@ proxy. Whether a covered cell currently passes or fails is a separate, live conc cd tests/e2e && PYTHONPATH=. python -m coverage_registry.collector ``` -Use `--format prometheus` or `--format json` for CI jobs that publish coverage to -Grafana. +Use `--format loki` after the e2e pytest run in the same Kubernetes job/pod to print +structured stdout lines for Loki: + +``` +cd tests/e2e && PYTHONPATH=. python -m coverage_registry.collector --format loki --strict +``` + +This emits exactly one `COVERAGE_TOTAL` line and one `COVERAGE_MODULE` line per module +in `MODULE_ORDER`, in that order. Loki uses log-safe `module=` labels from +`LOKI_MODULE_LABELS` (`core_llms`, `management_ui`, etc.) so existing JSON and +Prometheus consumers keep their human-readable module names unchanged. The headline is overall coverage. The collector also lists markers that point at ids not in the registry, so a typo or an unenumerated behavior surfaces instead of being diff --git a/tests/e2e/coverage_registry/collector.py b/tests/e2e/coverage_registry/collector.py index 3b577106605..f6e59ca4a88 100644 --- a/tests/e2e/coverage_registry/collector.py +++ b/tests/e2e/coverage_registry/collector.py @@ -21,7 +21,7 @@ from pathlib import Path import pytest from .registry import load_registry -from .schema import MODULE_ORDER, Cell, Tier, dashboard_module +from .schema import MODULE_ORDER, Cell, Tier, dashboard_module, loki_module_label E2E_DIR = Path(__file__).resolve().parent.parent @@ -239,13 +239,31 @@ def render_prometheus(report: CoverageReport) -> str: return "\n".join(lines) +def render_loki(report: CoverageReport) -> str: + lines = [ + ( + f"COVERAGE_TOTAL percent={report.coverage_percent:.1f} " + f"covered={report.covered} total={report.total}" + ) + ] + lines.extend( + ( + f"COVERAGE_MODULE module={loki_module_label(module.module)} " + f"percent={module.coverage_percent:.1f} " + f"covered={module.covered} total={module.total}" + ) + for module in report.modules + ) + return "\n".join(lines) + + def main() -> int: parser = ArgumentParser() parser.add_argument( "--format", - choices=("text", "json", "prometheus"), + choices=("text", "json", "prometheus", "loki"), default="text", - help="Output format. Use prometheus or json for Grafana ingestion jobs.", + help="Output format. Use loki for structured stdout lines in the e2e job.", ) parser.add_argument( "--strict", @@ -265,9 +283,8 @@ def main() -> int: "text": render, "json": render_json, "prometheus": render_prometheus, - }[ - args.format - ](report) + "loki": render_loki, + }[args.format](report) print(output) # noqa: T201 # CLI entrypoint output if args.strict and report.orphan_markers: return 1 diff --git a/tests/e2e/coverage_registry/schema.py b/tests/e2e/coverage_registry/schema.py index 2e2a00e78ba..bb27fbf0ea0 100644 --- a/tests/e2e/coverage_registry/schema.py +++ b/tests/e2e/coverage_registry/schema.py @@ -157,6 +157,16 @@ MODULE_ORDER: tuple[str, ...] = ( "Other", ) +LOKI_MODULE_LABELS: dict[str, str] = { + "Core LLMs": "core_llms", + "Non-Core LLMs": "non_core_llms", + "MCPs": "mcp", + "Management/UI": "management_ui", + "Reliability & Performance": "reliability_performance", + "Logging & Guardrails": "logging_guardrails", + "Other": "other", +} + def dashboard_module(cell: Cell) -> str: """Return the Grafana/reporting module for a registry cell.""" @@ -165,3 +175,8 @@ def dashboard_module(cell: Cell) -> str: return "Core LLMs" return "Non-Core LLMs" return PREFIX_ROLLUP[cell.module] + + +def loki_module_label(module: str) -> str: + """Return the log-safe Loki label for a dashboard module.""" + return LOKI_MODULE_LABELS[module] diff --git a/tests/e2e/coverage_registry/test_collector.py b/tests/e2e/coverage_registry/test_collector.py index 355bc52730d..079ee215866 100644 --- a/tests/e2e/coverage_registry/test_collector.py +++ b/tests/e2e/coverage_registry/test_collector.py @@ -15,6 +15,7 @@ from coverage_registry.collector import ( compute_coverage, render, render_json, + render_loki, render_prometheus, ) from coverage_registry.registry import load_registry @@ -24,6 +25,7 @@ from coverage_registry.schema import ( LlmEndpoint, LoggingCell, Tier, + loki_module_label, ) @@ -149,6 +151,30 @@ def test_prometheus_render_exposes_module_coverage_timeseries() -> None: assert "litellm_e2e_coverage_orphan_markers 0" in metrics +def test_loki_render_exposes_exact_stdout_lines_for_loki() -> None: + report = compute_coverage( + (_llm("llm.chat", Tier.P0), _llm("llm.batches", Tier.P0, "batches")), + frozenset({"llm.chat"}), + ) + + lines = render_loki(report).splitlines() + + assert len(lines) == 1 + len(report.modules) + assert lines[0] == "COVERAGE_TOTAL percent=50.0 covered=1 total=2" + assert ( + lines[1] == "COVERAGE_MODULE module=core_llms percent=100.0 covered=1 total=1" + ) + assert ( + lines[2] == "COVERAGE_MODULE module=non_core_llms percent=0.0 covered=0 total=1" + ) + assert [line.split("module=", 1)[1].split(" ", 1)[0] for line in lines[1:]] == [ + loki_module_label(module.module) for module in report.modules + ] + assert all( + " " not in line.split("module=", 1)[1].split(" ", 1)[0] for line in lines[1:] + ) + + def test_real_registry_loads_and_ids_are_unique() -> None: cells = load_registry() ids = [c.id for c in cells] diff --git a/tests/e2e/docker-compose.yml b/tests/e2e/docker-compose.yml index cdf5d6cbf6f..195badc5285 100644 --- a/tests/e2e/docker-compose.yml +++ b/tests/e2e/docker-compose.yml @@ -25,6 +25,10 @@ configs: fallbacks: - gemini-2.5-flash: ["gpt-5.5", "claude-haiku-4-5"] + finetune_settings: + - custom_llm_provider: openai + api_key: os.environ/OPENAI_API_KEY + model_list: - model_name: gpt-5.5 litellm_params: diff --git a/tests/e2e/e2e_gateway.py b/tests/e2e/e2e_gateway.py index 055d06d1c79..05f83ecc085 100644 --- a/tests/e2e/e2e_gateway.py +++ b/tests/e2e/e2e_gateway.py @@ -28,6 +28,9 @@ from models import ( CustomerDeleteBody, EmbedBody, EmbedResponse, + FileListResponse, + FineTuningJobsParams, + FineTuningJobsResponse, KeyDeleteBody, KeyGenerateBody, KeyGenerateResponse, @@ -42,6 +45,7 @@ from models import ( ModelMode, ModelNewBody, ModelNewResponse, + ModelsListResponse, OcrBody, OcrResponse, SpendLogRow, @@ -119,27 +123,83 @@ class Gateway: ) ).data + def list_files(self, key: str) -> Result[FileListResponse]: + return self.transport.get( + "/v1/files", + headers=self.transport.bearer(key), + params=NoBody(), + response_type=FileListResponse, + ) + + def list_fine_tuning_jobs( + self, key: str, params: FineTuningJobsParams + ) -> Result[FineTuningJobsResponse]: + return self.transport.get( + "/v1/fine_tuning/jobs", + headers=self.transport.bearer(key), + params=params, + response_type=FineTuningJobsResponse, + ) + def create_model( self, model_name: str, litellm_params: LiteLLMParamsBody, mode: ModelMode | None = None, ) -> str: - """Register a deployment under `model_name` (id == model_name) and return the - model_id. add_deployment runs synchronously in /model/new, so the model is - callable as soon as this returns.""" - return unwrap( + """Register a deployment under `model_name` and return its proxy-assigned + model_id, once the model is actually servable on the data plane. + + /model/new is a control-plane route; in a split control/data-plane + deployment the gateway (data plane, which serves /chat, /ocr, ...) only + picks the new model up on its next DB reload, so a call issued the instant + this returns can race the reload and 400 with "Invalid model name passed". + We therefore poll the data-plane /v1/models until the model appears before + handing back, so callers can invoke it immediately. In the monolithic case + it is already present on the first poll, so this adds one request.""" + model_id = unwrap( self.transport.post( "/model/new", headers=self.transport.master, json=ModelNewBody( model_name=model_name, litellm_params=litellm_params, - model_info=ModelInfoBody(id=model_name, mode=mode), + model_info=ModelInfoBody(mode=mode), ), response_type=ModelNewResponse, ) ).model_id + self._await_model_servable(model_name) + return model_id + + def _await_model_servable(self, model_name: str) -> None: + """Block until the data plane lists `model_name`, or fail loudly if it does + not within poll_timeout (a real propagation/config problem, surfaced here + instead of as a downstream "Invalid model name passed").""" + deadline = time.monotonic() + self.poll_timeout + last_result: Result[ModelsListResponse] | None = None + while time.monotonic() < deadline: + last_result = self.transport.get( + "/v1/models", + headers=self.transport.master, + params=NoBody(), + response_type=ModelsListResponse, + ) + if isinstance(last_result, Success) and any( + entry.id == model_name for entry in last_result.data.data + ): + return + time.sleep(self.poll_interval) + last_error = ( + f"; last /v1/models poll did not succeed: {last_result}" + if last_result is not None and not isinstance(last_result, Success) + else "" + ) + raise AssertionError( + f"model {model_name!r} was created but never became servable on the data " + f"plane within {self.poll_timeout}s of /model/new (control/data-plane " + f"propagation or STORE_MODEL_IN_DB reload issue){last_error}" + ) def delete_model(self, model_id: str) -> None: result = self.transport.post( diff --git a/tests/e2e/llm_translation/realtime/REALTIME_COVERAGE_MATRIX.md b/tests/e2e/llm_translation/realtime/REALTIME_COVERAGE_MATRIX.md new file mode 100644 index 00000000000..ffec7176f10 --- /dev/null +++ b/tests/e2e/llm_translation/realtime/REALTIME_COVERAGE_MATRIX.md @@ -0,0 +1,66 @@ +# Realtime e2e coverage + +Live tests for the proxy realtime websocket endpoint (`/v1/realtime`). One +GA-speaking websocket client drives every provider; the proxy normalizes each +provider's stream into the OpenAI GA event schema, so the same assertions hold +across providers and only the model alias changes. + +## What is asserted + +For each configured provider, `test_text_conversation` checks the session +lifecycle (`session.created`, then `session.update` echoed by `session.updated`), +the canonical response sequence (`response.created`, `response.output_item.added`, +through `response.done`), that the streamed deltas reconstruct a non-empty +transcript, and that `response.done` carries normalized usage. + +`test_tool_call_round_trip` checks the full tool path: the model emits a +normalized `response.function_call_arguments.done` with valid JSON arguments and +a matching `function_call` output item, the test sends a `function_call_output` +back, and the follow-up response incorporates the result (the temperature 72 +appears). + +`test_realtime_pipecat_e2e` is a realism layer that drives the same providers +through pipecat's GA `OpenAIRealtimeLLMService` (base_url pointed at the proxy) +rather than speaking the protocol by hand. Its assertions are coarse (the tool +callback fired, assistant text was produced); the raw-websocket suite is the +source of truth. It skips unless `pipecat-ai` is installed +(`uv pip install "pipecat-ai[openai]"`). + +## Provisioning + +The suite registers every provider's realtime deployment through `/model/new` at +session start (the `realtime_models` fixture) and deletes them on teardown, so it +never depends on a static or misconfigured gateway `model_list`. Each deployment +is created with `model_info.mode: realtime` and marker-unique names, and its +`litellm_params` point the credentials at `os.environ/*` refs the gateway resolves +at call time. The provider table below is the source of truth; edit `PROVIDERS` in +`realtime_client.py` to change a model or add one. + +| provider | model alias | upstream model | +|----------|-------------|----------------| +| openai | `openai-realtime` | `openai/gpt-realtime-2` | +| azure | `azure-realtime` | `azure/gpt-realtime-2` (GA protocol) | +| gemini | `gemini-realtime` | `gemini/gemini-3.1-flash-live-preview` | +| vertex_ai | `vertex-realtime` | `vertex_ai/gemini-live-2.5-flash-preview-native-audio-09-2025` | + +Bedrock and xai (`xai/grok-4-1-fast`) are supported by the proxy but +kept commented out in `PROVIDERS` until they pass end-to-end here; re-enable them by +uncommenting their entry. + +Every provider is provisioned and asserted; the suite never skips a provider. Per +`tests/e2e/CLAUDE.md` the only sanctioned skip is the whole-suite proxy-liveness +skip, so a provider whose credentials or upstream realtime model are missing on the +gateway is a hard failure, not a skip. Give the gateway each provider's credentials +to turn its tests green. + +## Running + +Start a proxy with the provider keys set in its environment (the suite registers +the deployments itself), then + +``` +uv run pytest tests/e2e/llm_translation/realtime/ -v +``` + +The whole suite skips only when no proxy answers `GET /health/liveliness` at +`LITELLM_PROXY_URL` (default `http://localhost:4000`). diff --git a/tests/e2e/llm_translation/realtime/conftest.py b/tests/e2e/llm_translation/realtime/conftest.py new file mode 100644 index 00000000000..15cd789664e --- /dev/null +++ b/tests/e2e/llm_translation/realtime/conftest.py @@ -0,0 +1,37 @@ +"""Realtime suite's `client` and `realtime_models` fixtures. + +The shared lifecycle (resources/scoped_key), proxy liveness skip, and e2e marker +live in the parent tests/e2e/conftest.py. RealtimeClient holds the shared Gateway, +so the `resources` fixture cleans up keys this suite creates. + +`realtime_models` registers every provider's realtime deployment through /model/new +at session start and deletes them at teardown, so the suite provisions the models it +uses through the management endpoints instead of depending on a static (or +misconfigured) gateway model_list. +""" + +from collections.abc import Iterator + +import pytest + +from realtime_client import PROVIDERS, RealtimeClient, build_client + + +@pytest.fixture(scope="session") +def client() -> RealtimeClient: + return build_client() + + +@pytest.fixture(scope="session") +def realtime_models(client: RealtimeClient) -> Iterator[dict[str, str]]: + """Provision each provider's realtime deployment via /model/new and yield a + provider-id -> model-name map the tests connect with; delete them on teardown. + Every provider is provisioned (never skipped): a provider whose credentials or + upstream model are missing on the gateway hard-fails its test, per the suite's + fail-on-behavior contract in tests/e2e/CLAUDE.md.""" + records = tuple((provider.id, *client.provision(provider)) for provider in PROVIDERS) + try: + yield {provider_id: model_name for provider_id, model_name, _ in records} + finally: + for _, _, model_id in records: + client.gateway.delete_model(model_id) diff --git a/tests/e2e/realtime/fixtures/weather_question_24k.wav b/tests/e2e/llm_translation/realtime/fixtures/weather_question_24k.wav similarity index 100% rename from tests/e2e/realtime/fixtures/weather_question_24k.wav rename to tests/e2e/llm_translation/realtime/fixtures/weather_question_24k.wav diff --git a/tests/e2e/realtime/pipecat_service.py b/tests/e2e/llm_translation/realtime/pipecat_service.py similarity index 100% rename from tests/e2e/realtime/pipecat_service.py rename to tests/e2e/llm_translation/realtime/pipecat_service.py diff --git a/tests/e2e/realtime/realtime_client.py b/tests/e2e/llm_translation/realtime/realtime_client.py similarity index 69% rename from tests/e2e/realtime/realtime_client.py rename to tests/e2e/llm_translation/realtime/realtime_client.py index 07dfd76108b..51052abe227 100644 --- a/tests/e2e/realtime/realtime_client.py +++ b/tests/e2e/llm_translation/realtime/realtime_client.py @@ -11,19 +11,19 @@ models, matching the suite's no-raw-dicts rule. from __future__ import annotations import time -from collections.abc import Generator +from collections.abc import Generator, Mapping from contextlib import contextmanager from dataclasses import dataclass from typing import Any, TypeVar from urllib.parse import urlencode -import pytest from pydantic import BaseModel, ConfigDict from websockets.sync.client import connect from websockets.sync.connection import Connection -from e2e_config import PROXY_BASE_URL +from e2e_config import PROXY_BASE_URL, unique_marker from e2e_gateway import Gateway, build_gateway +from models import LiteLLMParamsBody _M = TypeVar("_M", bound=BaseModel) @@ -41,25 +41,76 @@ def realtime_ws_url(model: str) -> str: @dataclass(frozen=True, slots=True) class RealtimeProvider: + """A realtime provider the suite exercises. `litellm_params` is the deployment + the suite registers through /model/new (the gateway resolves the os.environ/* + credential refs), so the suite is self-contained and never depends on a static + gateway model_list. Every provider here is provisioned and asserted: per + tests/e2e/CLAUDE.md the suite never skips a provider, so a provider whose + credentials or upstream realtime model are missing on the gateway is a hard + failure, not a skip.""" + id: str - model: str + alias: str + litellm_params: LiteLLMParamsBody PROVIDERS = ( - RealtimeProvider("openai", "openai-realtime"), - RealtimeProvider("azure", "azure-realtime"), - RealtimeProvider("gemini", "gemini-realtime"), - RealtimeProvider("vertex_ai", "vertex-realtime"), - # RealtimeProvider("bedrock", "bedrock-realtime"), # TODO: Enable this when Bedrock is passing - RealtimeProvider("xai", "xai-realtime"), + RealtimeProvider( + "openai", + "openai-realtime", + LiteLLMParamsBody( + model="openai/gpt-realtime-2", + api_key="os.environ/OPENAI_API_KEY", + ), + ), + RealtimeProvider( + "azure", + "azure-realtime", + LiteLLMParamsBody( + model="azure/gpt-realtime", + api_key="os.environ/AZURE_API_KEY", + api_version="2025-08-28", + realtime_protocol="GA", + ), + ), + RealtimeProvider( + "gemini", + "gemini-realtime", + LiteLLMParamsBody( + model="gemini/gemini-3.1-flash-live-preview", + api_key="os.environ/GEMINI_API_KEY", + ), + ), + RealtimeProvider( + "vertex_ai", + "vertex-realtime", + LiteLLMParamsBody( + model="vertex_ai/gemini-live-2.5-flash-preview-native-audio-09-2025", + vertex_location="us-central1", + vertex_credentials="os.environ/VERTEXAI_CREDENTIALS", + ), + ), + # RealtimeProvider("bedrock", "bedrock-realtime", ...) # TODO: Enable when Bedrock is passing + # RealtimeProvider( + # "xai", + # "xai-realtime", + # LiteLLMParamsBody( + # model="xai/grok-4-1-fast", + # api_key="os.environ/XAI_API_KEY", + # ), + # ), # TODO: Enable once xai Grok Voice realtime is passing end-to-end here ) -def skip_if_unconfigured( - provider: RealtimeProvider, configured: frozenset[str] -) -> None: - if provider.model not in configured: - pytest.skip(f"{provider.model} not configured on proxy") +def realtime_model(provider: RealtimeProvider, provisioned: Mapping[str, str]) -> str: + """Return the provisioned deployment name for this provider. Every provider in + PROVIDERS is provisioned at session start, so a missing entry is a harness bug, + never an environment skip - the suite hard-fails instead (see tests/e2e/CLAUDE.md).""" + model = provisioned.get(provider.id) + assert model is not None, ( + f"{provider.id} was not provisioned; the realtime_models fixture is broken" + ) + return model # ---- sent events ------------------------------------------------------- @@ -280,12 +331,17 @@ class RealtimeSession: class RealtimeClient: gateway: Gateway - def configured_models(self) -> frozenset[str]: - return frozenset( - entry.model_name - for entry in self.gateway.model_info() - if entry.model_info.mode == "realtime" + def provision(self, provider: RealtimeProvider) -> tuple[str, str]: + """Register this provider's realtime deployment through /model/new and return + (model_name, model_id). The name is marker-unique so it never collides with a + same-named deployment already on the shared proxy, and mode=realtime makes it + show up as a realtime model on /model/info. add_deployment runs synchronously, + so the deployment is connectable as soon as this returns.""" + model_name = f"{provider.alias}-{unique_marker()}" + model_id = self.gateway.create_model( + model_name, provider.litellm_params, mode="realtime" ) + return model_name, model_id @contextmanager def connect( diff --git a/tests/e2e/realtime/test_realtime_e2e.py b/tests/e2e/llm_translation/realtime/test_realtime_e2e.py similarity index 92% rename from tests/e2e/realtime/test_realtime_e2e.py rename to tests/e2e/llm_translation/realtime/test_realtime_e2e.py index 01356900141..6aaffdd208e 100644 --- a/tests/e2e/realtime/test_realtime_e2e.py +++ b/tests/e2e/llm_translation/realtime/test_realtime_e2e.py @@ -31,7 +31,7 @@ from realtime_client import ( SessionUpdate, function_call_item, parse_last, - skip_if_unconfigured, + realtime_model, transcript, user_message, ) @@ -62,12 +62,12 @@ class WeatherResult(BaseModel): def test_text_conversation( client: RealtimeClient, scoped_key: str, - configured_models: frozenset[str], + realtime_models: dict[str, str], provider: RealtimeProvider, ) -> None: - skip_if_unconfigured(provider, configured_models) + model = realtime_model(provider, realtime_models) - with client.connect(key=scoped_key, model=provider.model) as session: + with client.connect(key=scoped_key, model=model) as session: created = session.collect_until("session.created", timeout=20) assert created[-1].type == "session.created" @@ -99,12 +99,12 @@ def test_text_conversation( def test_tool_call_round_trip( client: RealtimeClient, scoped_key: str, - configured_models: frozenset[str], + realtime_models: dict[str, str], provider: RealtimeProvider, ) -> None: - skip_if_unconfigured(provider, configured_models) + model = realtime_model(provider, realtime_models) - with client.connect(key=scoped_key, model=provider.model) as session: + with client.connect(key=scoped_key, model=model) as session: session.collect_until("session.created", timeout=20) session.send( SessionUpdate( diff --git a/tests/e2e/realtime/test_realtime_pipecat_audio_e2e.py b/tests/e2e/llm_translation/realtime/test_realtime_pipecat_audio_e2e.py similarity index 95% rename from tests/e2e/realtime/test_realtime_pipecat_audio_e2e.py rename to tests/e2e/llm_translation/realtime/test_realtime_pipecat_audio_e2e.py index d9d7c744f66..935d7528c1b 100644 --- a/tests/e2e/realtime/test_realtime_pipecat_audio_e2e.py +++ b/tests/e2e/llm_translation/realtime/test_realtime_pipecat_audio_e2e.py @@ -31,7 +31,7 @@ from realtime_client import ( PROVIDERS, RealtimeProvider, _ws_base_url, - skip_if_unconfigured, + realtime_model, ) pytestmark = pytest.mark.e2e @@ -96,7 +96,7 @@ SERVER_VAD_SETTINGS = rt_events.SessionProperties( noise_reduction=rt_events.InputAudioNoiseReduction(type="near_field"), turn_detection=rt_events.TurnDetection( type="server_vad", - threshold=0.8, + threshold=0.5, prefix_padding_ms=300, silence_duration_ms=700, ), @@ -189,13 +189,13 @@ async def _run_pipeline( @pytest.mark.parametrize("provider", PROVIDER_PARAMS) def test_pipecat_server_vad( scoped_key: str, - configured_models: frozenset[str], + realtime_models: dict[str, str], provider: RealtimeProvider, ) -> None: """Session is configured with server-VAD; bot must respond to a text prompt.""" - skip_if_unconfigured(provider, configured_models) + model = realtime_model(provider, realtime_models) - tool_called, got_text, _ = asyncio.run(_run_pipeline(scoped_key, provider.model)) + tool_called, got_text, _ = asyncio.run(_run_pipeline(scoped_key, model)) assert tool_called, "get_weather tool was not invoked" assert got_text, "no assistant text frames produced" @@ -204,16 +204,16 @@ def test_pipecat_server_vad( @pytest.mark.parametrize("provider", PROVIDER_PARAMS) def test_pipecat_audio_output( scoped_key: str, - configured_models: frozenset[str], + realtime_models: dict[str, str], provider: RealtimeProvider, ) -> None: """Bot must produce at least one non-empty TTS audio frame.""" - skip_if_unconfigured(provider, configured_models) + model = realtime_model(provider, realtime_models) _, got_text, audio_bytes = asyncio.run( _run_pipeline( scoped_key, - provider.model, + model, prompt="Say hello in one short sentence.", timeout=30.0, ) @@ -328,7 +328,7 @@ async def _run_audio_input_pipeline( @pytest.mark.parametrize("provider", PROVIDER_PARAMS) def test_pipecat_server_vad_audio_input( scoped_key: str, - configured_models: frozenset[str], + realtime_models: dict[str, str], provider: RealtimeProvider, ) -> None: """Stream a real PCM16 WAV fixture; server VAD must detect speech end and respond. @@ -337,12 +337,11 @@ def test_pipecat_server_vad_audio_input( → server-VAD turn detection → response.create (auto) → assistant reply. No LLMRunFrame is sent — the response must be triggered entirely by VAD. """ - if not WEATHER_WAV.exists(): - pytest.skip(f"audio fixture not found: {WEATHER_WAV}") - skip_if_unconfigured(provider, configured_models) + assert WEATHER_WAV.exists(), f"audio fixture not found: {WEATHER_WAV}" + model = realtime_model(provider, realtime_models) got_text, audio_bytes = asyncio.run( - _run_audio_input_pipeline(scoped_key, provider.model) + _run_audio_input_pipeline(scoped_key, model) ) assert got_text, "server VAD did not trigger a response (no assistant text)" diff --git a/tests/e2e/realtime/test_realtime_pipecat_e2e.py b/tests/e2e/llm_translation/realtime/test_realtime_pipecat_e2e.py similarity index 97% rename from tests/e2e/realtime/test_realtime_pipecat_e2e.py rename to tests/e2e/llm_translation/realtime/test_realtime_pipecat_e2e.py index 1068c54fdec..799958ef4e3 100644 --- a/tests/e2e/realtime/test_realtime_pipecat_e2e.py +++ b/tests/e2e/llm_translation/realtime/test_realtime_pipecat_e2e.py @@ -29,7 +29,7 @@ from realtime_client import ( PROVIDERS, RealtimeProvider, _ws_base_url, - skip_if_unconfigured, + realtime_model, ) pytestmark = pytest.mark.e2e @@ -123,12 +123,12 @@ async def _run_pipeline(key: str, model: str) -> tuple[bool, bool]: @pytest.mark.parametrize("provider", PROVIDER_PARAMS) def test_pipecat_tool_smoke( scoped_key: str, - configured_models: frozenset[str], + realtime_models: dict[str, str], provider: RealtimeProvider, ) -> None: - skip_if_unconfigured(provider, configured_models) + model = realtime_model(provider, realtime_models) - tool_called, produced_text = asyncio.run(_run_pipeline(scoped_key, provider.model)) + tool_called, produced_text = asyncio.run(_run_pipeline(scoped_key, model)) assert tool_called, "pipecat did not invoke the get_weather callback" assert produced_text, "pipecat produced no assistant text frames" diff --git a/tests/e2e/llm_translation/test_cache_control.py b/tests/e2e/llm_translation/test_cache_control.py new file mode 100644 index 00000000000..fda05cdc950 --- /dev/null +++ b/tests/e2e/llm_translation/test_cache_control.py @@ -0,0 +1,163 @@ +"""Live e2e: provider-specific /chat/completions features take real effect. + +Each case asserts the feature actually happened, not just a 200. Coverage matrix +(register-on-demand deployments, deleted on teardown): + +- Bedrock (anthropic claude-haiku-4-5): prompt caching. A large cacheable prefix + marked with ``cache_control`` is sent twice; the second call must report + cache-read usage tokens > 0. service_tier is out of scope for Bedrock; AWS + Bedrock does not expose an OpenAI-style request service tier, so that cell is + intentionally not covered here. +- Vertex (gemini-2.5-flash): prompt caching via ``cache_control`` context + caching; the second identical call must report cached prompt tokens > 0. + +service_tier lives in test_provider_features_e2e.py. + +The provider-native cache_control request shape is not expressible with the +shared ``ChatBody`` (whose content is a plain string), so the cacheable body is +modelled locally with typed content blocks. +""" + +from __future__ import annotations + +import time + +import pytest +from pydantic import BaseModel + +from e2e_config import unique_marker +from e2e_http import Result, unwrap +from lifecycle import ResourceManager +from models import ChatResponse, LiteLLMParamsBody, Usage +from passthrough_client import PassthroughClient +import os + +pytestmark = pytest.mark.e2e + +BEDROCK_MODEL = "bedrock/us.anthropic.claude-haiku-4-5-20251001-v1:0" +VERTEX_MODEL = "vertex_ai/gemini-2.5-flash" + + +class CacheControl(BaseModel): + type: str = "ephemeral" + + +class TextBlock(BaseModel): + type: str = "text" + text: str + cache_control: CacheControl | None = None + + +class RichMessage(BaseModel): + role: str + content: list[TextBlock] + + +class CacheChatBody(BaseModel): + model: str + messages: list[RichMessage] + max_tokens: int = 64 + cache: dict[str, bool] = {"no-cache": True} + + +def _cacheable_prefix() -> str: + """A prefix long enough to clear provider minimum cacheable sizes (Haiku is + 2048 tokens), unique per run so the first call writes and the second reads.""" + marker = unique_marker() + body = " ".join( + f"Cacheable reference paragraph {index} for run {marker}." for index in range(600) + ) + return f"{body}\nEnd of reference material {marker}." + + +def _cached_read_tokens(usage: Usage | None) -> int: + """Cache-read tokens however the provider reports them: Anthropic-style + ``cache_read_input_tokens`` or OpenAI-style ``prompt_tokens_details.cached_tokens``.""" + if usage is None: + return 0 + if usage.cache_read_input_tokens: + return usage.cache_read_input_tokens + if usage.prompt_tokens_details and usage.prompt_tokens_details.cached_tokens: + return usage.prompt_tokens_details.cached_tokens + return 0 + + +def _cache_chat( + client: PassthroughClient, key: str, model: str, prefix: str +) -> Result[ChatResponse]: + body = CacheChatBody( + model=model, + messages=[ + RichMessage( + role="system", + content=[TextBlock(text=prefix, cache_control=CacheControl())], + ), + RichMessage(role="user", content=[TextBlock(text="Reply with one word.")]), + ], + ) + return client.gateway.transport.post( + "/chat/completions", + headers=client.gateway.transport.bearer(key), + json=body, + response_type=ChatResponse, + ) + + +def _assert_cache_read_on_second_call( + client: PassthroughClient, key: str, model: str +) -> None: + prefix = _cacheable_prefix() + + first = unwrap(_cache_chat(client, key, model, prefix)) + assert first.choices, f"{model}: first cache-priming call returned no choices: {first}" + + read_tokens = 0 + deadline = time.monotonic() + 30.0 + while time.monotonic() < deadline: + second = unwrap(_cache_chat(client, key, model, prefix)) + read_tokens = _cached_read_tokens(second.usage) + if read_tokens > 0: + break + time.sleep(3.0) + + assert read_tokens > 0, ( + f"{model}: second identical call reported no cache-read tokens " + f"({second.usage}); prompt caching did not take effect" + ) + + +class TestCacheControl: + @pytest.mark.covers( + "llm.chat_completions.bedrock_converse.prompt_cache_5m.nonstream.works", + exercised_on=[], + ) + def test_bedrock_prompt_caching_reads_cache( + self, client: PassthroughClient, resources: ResourceManager + ) -> None: + model = f"e2e-bedrock-cache-{unique_marker()}" + model_id = client.gateway.create_model( + model, + LiteLLMParamsBody(model=BEDROCK_MODEL, aws_region_name="us-east-1"), + ) + resources.defer(lambda: client.gateway.delete_model(model_id)) + _assert_cache_read_on_second_call(client, resources.key(), model) + + @pytest.mark.covers( + "llm.chat_completions.vertex.prompt_cache_5m.nonstream.works", + exercised_on=[], + ) + def test_vertex_prompt_caching_reads_cache( + self, client: PassthroughClient, resources: ResourceManager + ) -> None: + model = f"e2e-vertex-cache-{unique_marker()}" + model_id = client.gateway.create_model( + model, + LiteLLMParamsBody( + model=VERTEX_MODEL, + vertex_project=os.environ.get("VERTEXAI_PROJECT"), + vertex_location="us-central1", + vertex_credentials=os.environ.get("VERTEXAI_CREDENTIALS"), + ), + ) + resources.defer(lambda: client.gateway.delete_model(model_id)) + _assert_cache_read_on_second_call(client, resources.key(), model) diff --git a/tests/e2e/llm_translation/test_deepseek_reasoning_e2e.py b/tests/e2e/llm_translation/test_deepseek_reasoning_e2e.py index f8f229aa2a7..5adb8c24f9f 100644 --- a/tests/e2e/llm_translation/test_deepseek_reasoning_e2e.py +++ b/tests/e2e/llm_translation/test_deepseek_reasoning_e2e.py @@ -2,17 +2,14 @@ DeepSeek's reasoner defaults thinking ON and surfaces the chain as ``message.reasoning_content``. Two documented ways to disable it are -``reasoning_effort="none"`` and ``thinking={"type": "disabled"}``. Today the -DeepSeek param mapper (``litellm/llms/deepseek/chat/transformation.py`` -``map_openai_params``) drops both without forwarding any disable signal, so the -outbound body carries no ``thinking`` key and DeepSeek keeps thinking on; the -response still comes back with ``reasoning_content``. That is the product gap -tracked by LIT-3686 / GH #27453. +``reasoning_effort="none"`` and ``thinking={"type": "disabled"}``. The DeepSeek +param mapper (``litellm/llms/deepseek/chat/transformation.py`` +``map_openai_params``) forwards both as ``thinking={"type": "disabled"}`` so the +outbound body carries a real disable signal and ``deepseek-reasoner`` returns no +``reasoning_content``. This is the behavior tracked by LIT-3686 / GH #27453. The control case proves the model and path work (reasoning is returned when -nothing asks to disable it), so the two disable assertions are meaningful. Those -two are marked xfail(strict) until the mapper forwards a real disable signal; an -xpass then alerts that the fix landed. +nothing asks to disable it), so the two disable assertions are meaningful. Requires DEEPSEEK_API_KEY on the proxy (tests/e2e/.env). No skip gate: once the proxy is up, a failure here is real, per the suite's hard-fail contract. @@ -74,13 +71,6 @@ class TestDeepSeekReasoningDisable: f"disable param, so the disable assertions below can't be trusted: {response}" ) - @pytest.mark.xfail( - strict=True, - reason=( - "LIT-3686 / GH #27453: DeepSeek reasoning_effort='none' and " - "thinking type='disabled' are silently dropped; reasoning not disabled" - ), - ) def test_reasoning_effort_none_disables_reasoning( self, client: PassthroughClient, resources: ResourceManager ) -> None: @@ -103,13 +93,6 @@ class TestDeepSeekReasoningDisable: f"is still present: {response}" ) - @pytest.mark.xfail( - strict=True, - reason=( - "LIT-3686 / GH #27453: DeepSeek reasoning_effort='none' and " - "thinking type='disabled' are silently dropped; reasoning not disabled" - ), - ) def test_thinking_disabled_disables_reasoning( self, client: PassthroughClient, resources: ResourceManager ) -> None: diff --git a/tests/e2e/llm_translation/test_ocr_rust_e2e.py b/tests/e2e/llm_translation/test_ocr_rust_e2e.py index 361bb5126a7..921010e5eae 100644 --- a/tests/e2e/llm_translation/test_ocr_rust_e2e.py +++ b/tests/e2e/llm_translation/test_ocr_rust_e2e.py @@ -86,16 +86,18 @@ class AzureDocIntelligenceOcr: @dataclass(frozen=True, slots=True) class VertexOcr: + """Vertex AI OCR (Mistral publisher). Only the location (not a secret) is set; + the project and credentials are left unset so the gateway resolves VERTEXAI_PROJECT + and VERTEXAI_CREDENTIALS from its own environment by name, keeping every secret on + the gateway like the azure_ai cases above. This is deliberate: the OCR path reads + vertex_project verbatim from litellm_params and never unwraps an `os.environ/*` + ref, so passing one would put the literal string in the request URL.""" + model: str location: str def litellm_params(self) -> LiteLLMParamsBody: - return LiteLLMParamsBody( - model=self.model, - vertex_project="os.environ/VERTEXAI_PROJECT", - vertex_location=self.location, - vertex_credentials="os.environ/VERTEXAI_CREDENTIALS", - ) + return LiteLLMParamsBody(model=self.model, vertex_location=self.location) @dataclass(frozen=True, slots=True) @@ -113,7 +115,7 @@ RUST_OCR_CASES: tuple[_OcrCase, ...] = ( ), _OcrCase( "azure-ai", - AzureAiOcr("azure_ai/mistral-document-ai-2505"), + AzureAiOcr("azure_ai/mistral-document-ai-2512"), OcrDocument(type="document_url", document_url=TEST_PDF_URL), ), _OcrCase( @@ -126,11 +128,6 @@ RUST_OCR_CASES: tuple[_OcrCase, ...] = ( VertexOcr("vertex_ai/mistral-ocr-2505", "us-central1"), OcrDocument(type="document_url", document_url=TEST_PDF_URL), ), - _OcrCase( - "vertex-deepseek", - VertexOcr("vertex_ai/deepseek-ocr-maas", "global"), - OcrDocument(type="image_url", image_url=TEST_IMAGE_URL), - ), ) _CASE_IDS = tuple(case.suffix for case in RUST_OCR_CASES) @@ -155,3 +152,5 @@ class TestRustOcrGateway: response = unwrap(endpoints_client.gateway.ocr(key, OcrBody(model=model, document=case.document))) _assert_ocr_document(response) + + diff --git a/tests/e2e/llm_translation/test_provider_features_e2e.py b/tests/e2e/llm_translation/test_provider_features_e2e.py index cf05a4306b4..822a1d8d4d5 100644 --- a/tests/e2e/llm_translation/test_provider_features_e2e.py +++ b/tests/e2e/llm_translation/test_provider_features_e2e.py @@ -1,78 +1,32 @@ -"""Live e2e for model-specific request features: service_tier and prompt caching. +"""Live e2e for model-specific request features: service_tier. Each case asserts the feature took effect, not just a 200. service_tier is an OpenAI concept. The proxy forwards it and the provider echoes -the tier back on the response, so sending a non-default tier ("flex") and reading -it back off ``service_tier`` proves the param was honored end to end; litellm's own -default injection would report "default", so a "flex" echo can only come from the -request being forwarded. Bedrock and Vertex do not accept service_tier, so that -cell is OpenAI-only by design. +the tier back on the response, so sending a non-default tier ("priority") and +reading it back off ``service_tier`` proves the param was honored end to end; +litellm's own default injection (and service_tier="auto") both report "default", +so a "priority" echo can only come from the request being forwarded. "flex" is +avoided here because it is capacity-constrained and returns a transient 429 when +flex resources are unavailable. Bedrock and Vertex do not accept service_tier, so +that cell is OpenAI-only by design. -Prompt caching is asserted through provider prompt-cache usage tokens. The -deterministic path is explicit ``cache_control`` on an Anthropic-family model -(here Bedrock's Claude): a large cacheable prefix is sent twice and the second -call must report ``cache_read_input_tokens > 0``. OpenAI and Gemini only offer -implicit automatic caching, which does not deterministically produce a cache read -within a test window (verified: repeated >3k-token prompts kept -``prompt_tokens_details.cached_tokens`` at 0), so those caching cells are out of -scope here and covered only by the explicit-cache-control Bedrock case. +Prompt caching lives in test_cache_control.py. """ from __future__ import annotations import pytest -from pydantic import BaseModel from e2e_config import unique_marker from e2e_http import unwrap from lifecycle import ResourceManager -from models import ChatBody, ChatMessage, ChatResponse, LiteLLMParamsBody +from models import ChatBody, ChatMessage, LiteLLMParamsBody from passthrough_client import PassthroughClient pytestmark = pytest.mark.e2e -SERVICE_TIER = "flex" -CACHE_MIN_READ_TOKENS = 1 - - -class CacheControl(BaseModel): - type: str = "ephemeral" - - -class CacheTextBlock(BaseModel): - type: str = "text" - text: str - cache_control: CacheControl | None = None - - -class RichMessage(BaseModel): - role: str - content: list[CacheTextBlock] - - -class CacheChatBody(BaseModel): - model: str - messages: list[RichMessage] - max_tokens: int - - -def cacheable_prefix() -> str: - return ( - "You are a policy compliance auditor. The following corpus is the immutable " - "reference the assistant must consult on every turn. " - ) + ("Clause: obey all safety, formatting, and citation rules exactly. " * 400) - - -def post_chat(client: PassthroughClient, key: str, body: BaseModel) -> ChatResponse: - return unwrap( - client.gateway.transport.post( - "/chat/completions", - headers=client.gateway.transport.bearer(key), - json=body, - response_type=ChatResponse, - ) - ) +SERVICE_TIER = "priority" class TestServiceTier: @@ -107,50 +61,3 @@ class TestServiceTier: f"service_tier not honored: sent {SERVICE_TIER!r}, response reported " f"{response.service_tier!r} ({response})" ) - - -class TestPromptCaching: - @pytest.mark.covers( - "llm.chat_completions.bedrock_converse.prompt_cache_5m.nonstream.works", - exercised_on=[], - ) - def test_bedrock_cache_control_produces_cache_read( - self, client: PassthroughClient, resources: ResourceManager - ) -> None: - model = f"e2e-bedrock-cache-{unique_marker()}" - model_id = client.gateway.create_model( - model, - LiteLLMParamsBody( - model="bedrock/us.anthropic.claude-haiku-4-5-20251001-v1:0", - aws_region_name="us-east-1", - ), - ) - resources.defer(lambda: client.gateway.delete_model(model_id)) - key = resources.key() - - body = CacheChatBody( - model=model, - max_tokens=32, - messages=[ - RichMessage( - role="user", - content=[ - CacheTextBlock( - text=cacheable_prefix(), cache_control=CacheControl() - ), - CacheTextBlock(text="Answer in one word: acknowledged?"), - ], - ) - ], - ) - - first = post_chat(client, key, body) - assert first.usage is not None, f"first call reported no usage: {first}" - - second = post_chat(client, key, body) - assert second.usage is not None, f"second call reported no usage: {second}" - cache_read = second.usage.cache_read_input_tokens - assert cache_read is not None and cache_read >= CACHE_MIN_READ_TOKENS, ( - "second identical request did not read the prompt cache: " - f"cache_read_input_tokens={cache_read!r} (usage={second.usage})" - ) diff --git a/tests/e2e/models.py b/tests/e2e/models.py index 0490db286ea..38778034de9 100644 --- a/tests/e2e/models.py +++ b/tests/e2e/models.py @@ -376,6 +376,7 @@ class LiteLLMParamsBody(BaseModel): api_key: str | None = None api_base: str | None = None api_version: str | None = None + realtime_protocol: str | None = None aws_region_name: str | None = None vertex_project: str | None = None vertex_location: str | None = None @@ -394,7 +395,11 @@ ModelMode = Literal["batch", "realtime", "image_generation"] class ModelInfoBody(BaseModel): - id: str + # id is left unset so the proxy assigns a unique model_id per deployment. + # Pinning it to the model_name made re-registrations of a fixed-name model + # (e.g. the batch suite's openai-batch) collide on the model_id unique + # constraint when a prior run's teardown had not removed the row. + id: str | None = None mode: ModelMode | None = None @@ -410,6 +415,18 @@ class ModelNewResponse(BaseModel): model_id: str +class ModelListEntry(BaseModel): + id: str + + +class ModelsListResponse(BaseModel): + """GET /v1/models on the data plane: the deployments the gateway can actually + serve right now. Used to confirm a freshly created model has propagated from + the control plane before a test calls it.""" + + data: tuple[ModelListEntry, ...] = () + + class ModelDeleteBody(BaseModel): id: str diff --git a/tests/e2e/realtime/REALTIME_COVERAGE_MATRIX.md b/tests/e2e/realtime/REALTIME_COVERAGE_MATRIX.md deleted file mode 100644 index 8624475d0de..00000000000 --- a/tests/e2e/realtime/REALTIME_COVERAGE_MATRIX.md +++ /dev/null @@ -1,55 +0,0 @@ -# Realtime e2e coverage - -Live tests for the proxy realtime websocket endpoint (`/v1/realtime`). One -GA-speaking websocket client drives every provider; the proxy normalizes each -provider's stream into the OpenAI GA event schema, so the same assertions hold -across providers and only the model alias changes. - -## What is asserted - -For each configured provider, `test_text_conversation` checks the session -lifecycle (`session.created`, then `session.update` echoed by `session.updated`), -the canonical response sequence (`response.created`, `response.output_item.added`, -through `response.done`), that the streamed deltas reconstruct a non-empty -transcript, and that `response.done` carries normalized usage. - -`test_tool_call_round_trip` checks the full tool path: the model emits a -normalized `response.function_call_arguments.done` with valid JSON arguments and -a matching `function_call` output item, the test sends a `function_call_output` -back, and the follow-up response incorporates the result (the temperature 72 -appears). - -`test_realtime_pipecat_e2e` is a realism layer that drives the same providers -through pipecat's GA `OpenAIRealtimeLLMService` (base_url pointed at the proxy) -rather than speaking the protocol by hand. Its assertions are coarse (the tool -callback fired, assistant text was produced); the raw-websocket suite is the -source of truth. It skips unless `pipecat-ai` is installed -(`uv pip install "pipecat-ai[openai]"`). - -## Provider status - -| provider | model alias | status | -|----------|-------------|--------| -| openai | `openai-realtime` | covered (in gateway config) | -| gemini | `gemini-realtime` | covered (in gateway config; needs Gemini Live API access) | -| azure | `azure-realtime` | gap: add to gateway config + AZURE creds | -| vertex_ai | `vertex-realtime` | gap: add to gateway config + Vertex creds | -| bedrock | `bedrock-realtime` | gap: add to gateway config + AWS creds | -| xai | `xai-realtime` | gap: add to gateway config + XAI_API_KEY | - -A provider whose alias is not present in the proxy's `/model/info` skips (skip on -environment). To enable one, add a `model_info.mode: realtime` entry under that -alias to `tests/e2e/gateway/litellm-config.yml` and give the proxy the -provider's credentials; the test then runs with no code change. - -## Running - -Start a proxy with the gateway config and the provider keys set in its -environment, then - -``` -uv run pytest tests/e2e/realtime/ -v -``` - -Tests skip when no proxy answers `GET /health/liveliness` at `LITELLM_PROXY_URL` -(default `http://localhost:4000`). diff --git a/tests/e2e/realtime/conftest.py b/tests/e2e/realtime/conftest.py deleted file mode 100644 index 4a5c4837a1a..00000000000 --- a/tests/e2e/realtime/conftest.py +++ /dev/null @@ -1,20 +0,0 @@ -"""Realtime suite's `client` fixture. - -The shared lifecycle (resources/scoped_key), proxy liveness skip, and e2e marker -live in the parent tests/e2e/conftest.py. RealtimeClient holds the shared -Gateway, so the `resources` fixture cleans up keys this suite creates. -""" - -import pytest - -from realtime_client import RealtimeClient, build_client - - -@pytest.fixture(scope="session") -def client() -> RealtimeClient: - return build_client() - - -@pytest.fixture(scope="session") -def configured_models(client: RealtimeClient) -> frozenset[str]: - return client.configured_models() diff --git a/tests/e2e/test_e2e_gateway.py b/tests/e2e/test_e2e_gateway.py index a6dcc6112d6..9a9aa2fd2cc 100644 --- a/tests/e2e/test_e2e_gateway.py +++ b/tests/e2e/test_e2e_gateway.py @@ -10,6 +10,7 @@ signature drift fails here instead of in a live stage run. from dataclasses import dataclass, field +import pytest from pydantic import BaseModel from batches.batch_client import BatchClient @@ -21,26 +22,38 @@ from e2e_http import ( Result, StreamingResponse, Success, + UnknownApiError, ) from models import ( LiteLLMParamsBody, ModelDeleteBody, ModelNewBody, ModelNewResponse, + ModelsListResponse, ) @dataclass class _RecordingTransport: """Typed fake fulfilling the Transport protocol; records every post and - answers with a canned success so the test asserts on what was sent.""" + answers with a canned success so the test asserts on what was sent. + + `get("/v1/models")` reports a created model as servable only after + `servable_after_gets` polls, so a test can drive the data-plane wait in + create_model.""" posts: list[tuple[str, BaseModel]] = field(default_factory=list) + servable_after_gets: int = 0 + models_error: UnknownApiError | None = None + model_gets: int = 0 + _created: list[str] = field(default_factory=list) def post[R: BaseModel]( self, path: str, *, headers: BaseModel, json: BaseModel, response_type: type[R] ) -> Result[R]: self.posts.append((path, json)) + if path == "/model/new" and isinstance(json, ModelNewBody): + self._created.append(json.model_name) payload = ( {"model_id": "registered-id"} if response_type is ModelNewResponse else {} ) @@ -70,7 +83,15 @@ class _RecordingTransport: params: BaseModel, response_type: type[R], ) -> Result[R]: - raise AssertionError("get is not part of model management") + if path == "/v1/models" and response_type is ModelsListResponse: + self.model_gets += 1 + if self.models_error is not None: + return self.models_error + visible = self._created if self.model_gets > self.servable_after_gets else [] + return Success( + data=response_type.model_validate({"data": [{"id": name} for name in visible]}) + ) + raise AssertionError(f"unexpected get: {path}") def delete[R: BaseModel]( self, path: str, *, headers: BaseModel, json: BaseModel, response_type: type[R] @@ -106,7 +127,7 @@ class _RecordingTransport: def test_gateway_create_model_registers_deployment_and_returns_model_id() -> None: transport = _RecordingTransport() - gateway = Gateway(transport=transport) + gateway = Gateway(transport=transport, poll_interval=0.0) model_id = gateway.create_model( "e2e-test-model", LiteLLMParamsBody(model="openai/gpt-4o-mini") @@ -117,13 +138,48 @@ def test_gateway_create_model_registers_deployment_and_returns_model_id() -> Non assert path == "/model/new" assert isinstance(body, ModelNewBody) assert body.model_name == "e2e-test-model" - assert body.model_info.id == "e2e-test-model" + # No pinned model_id: the proxy assigns a unique one, so a fixed-name model + # re-registered after a failed teardown can't collide on the id constraint. + assert body.model_info.id is None assert body.model_info.mode is None + # It confirmed data-plane visibility before returning. + assert transport.model_gets >= 1 + + +def test_gateway_create_model_waits_until_servable_on_the_data_plane() -> None: + # The model shows up on /v1/models only on the third poll (simulating the + # gateway's delayed DB reload in a split deployment); create_model must keep + # polling instead of returning after /model/new. + transport = _RecordingTransport(servable_after_gets=2) + gateway = Gateway(transport=transport, poll_interval=0.0) + + gateway.create_model("e2e-late-model", LiteLLMParamsBody(model="openai/gpt-4o-mini")) + + assert transport.model_gets == 3 + + +def test_gateway_create_model_fails_loudly_when_never_servable() -> None: + transport = _RecordingTransport(servable_after_gets=10**9) + gateway = Gateway(transport=transport, poll_timeout=0.05, poll_interval=0.0) + + with pytest.raises(AssertionError, match="never became servable"): + gateway.create_model("e2e-ghost-model", LiteLLMParamsBody(model="openai/gpt-4o-mini")) + + +def test_gateway_create_model_surfaces_the_last_data_plane_error() -> None: + transport = _RecordingTransport( + models_error=UnknownApiError(status_code=503, body="data plane down") + ) + gateway = Gateway(transport=transport, poll_timeout=0.05, poll_interval=0.0) + + with pytest.raises(AssertionError, match="data plane down") as excinfo: + gateway.create_model("e2e-flaky-model", LiteLLMParamsBody(model="openai/gpt-4o-mini")) + assert "503" in str(excinfo.value) def test_batch_client_create_model_registers_a_batch_mode_deployment() -> None: transport = _RecordingTransport() - client = BatchClient(gateway=Gateway(transport=transport)) + client = BatchClient(gateway=Gateway(transport=transport, poll_interval=0.0)) model_id = client.create_model( "e2e-batch-model", LiteLLMParamsBody(model="openai/gpt-4o-mini") diff --git a/tests/enterprise/litellm_enterprise/proxy/guardrails/test_bedrock_apply_guardrail.py b/tests/enterprise/litellm_enterprise/proxy/guardrails/test_bedrock_apply_guardrail.py index 87c5e3bf2a9..f257b47404e 100644 --- a/tests/enterprise/litellm_enterprise/proxy/guardrails/test_bedrock_apply_guardrail.py +++ b/tests/enterprise/litellm_enterprise/proxy/guardrails/test_bedrock_apply_guardrail.py @@ -329,12 +329,13 @@ def test_bedrock_guardrail_filters_latest_user_message_when_enabled(): @pytest.mark.asyncio async def test_bedrock_apply_guardrail_blocked_with_disable_exception_on_block(): """ - Regression test for issue #20045: when disable_exception_on_block=True, - make_bedrock_api_request raises GuardrailInterventionNormalStringError. - apply_guardrail must let it propagate as-is so the proxy can handle it - properly instead of wrapping it in a generic Exception. + Regression test for LIT-4186: when disable_exception_on_block=True, a + Bedrock block raises ModifyResponseException. apply_guardrail must let it + propagate as-is so the endpoint handler (proxy_server.py) can turn it into + a 200 response with the block message as content, instead of the exception + surfacing as a bare 500. """ - from litellm.exceptions import GuardrailInterventionNormalStringError + from litellm.exceptions import ModifyResponseException guardrail = BedrockGuardrail( guardrail_name="test-bedrock-guard", @@ -346,18 +347,21 @@ async def test_bedrock_apply_guardrail_blocked_with_disable_exception_on_block() with patch.object( guardrail, "make_bedrock_api_request", new_callable=AsyncMock ) as mock_api: - mock_api.side_effect = GuardrailInterventionNormalStringError( - message="Sorry, your question in its current format is unable to be answered." + mock_api.side_effect = ModifyResponseException( + message="Sorry, your question in its current format is unable to be answered.", + model="bedrock-guardrail", + request_data={}, + guardrail_name="test-bedrock-guard", ) - with pytest.raises(GuardrailInterventionNormalStringError) as exc_info: + with pytest.raises(ModifyResponseException) as exc_info: await guardrail.apply_guardrail( inputs={"texts": ["harmful prompt content"]}, request_data={}, input_type="request", ) - assert "unable to be answered" in str(exc_info.value.message) + assert "unable to be answered" in exc_info.value.message @pytest.mark.asyncio diff --git a/tests/guardrails_tests/test_bedrock_guardrails.py b/tests/guardrails_tests/test_bedrock_guardrails.py index a23e89e576c..823ee05839f 100644 --- a/tests/guardrails_tests/test_bedrock_guardrails.py +++ b/tests/guardrails_tests/test_bedrock_guardrails.py @@ -1390,7 +1390,14 @@ async def test_bedrock_guardrail_disable_exception_on_block_non_streaming(): assert exception.status_code == 400 assert "Violated guardrail policy" in str(exception.detail) - # Test 2: disable_exception_on_block=True - should NOT raise exception + # Test 2: disable_exception_on_block=True - raises ModifyResponseException. + # LIT-4186: pre-fix, the native hook swallowed the block and set + # data["mock_response"], which was dead code (route_request already + # unpacked kwargs) so during_call let the model call proceed anyway. + # The correct contract is to raise ModifyResponseException so the endpoint + # handler returns a 200 with the block message as content. + from litellm.exceptions import ModifyResponseException + guardrail_disabled = BedrockGuardrail( guardrailIdentifier="test-guardrail", guardrailVersion="DRAFT", @@ -1402,20 +1409,13 @@ async def test_bedrock_guardrail_disable_exception_on_block_non_streaming(): ) as mock_post: mock_post.return_value = mock_bedrock_response - # Should NOT raise exception when disable_exception_on_block=True - try: - response = await guardrail_disabled.async_moderation_hook( + with pytest.raises(ModifyResponseException) as exc_info: + await guardrail_disabled.async_moderation_hook( data=request_data, user_api_key_dict=mock_user_api_key_dict, call_type="completion", ) - # Should succeed and return data (even though content was blocked) - assert response is not None - print("✅ No exception raised when disable_exception_on_block=True") - except Exception as e: - pytest.fail( - f"Should not raise exception when disable_exception_on_block=True, but got: {e}" - ) + assert exc_info.value.message == "I can't provide that information." @pytest.mark.asyncio @@ -1514,7 +1514,10 @@ async def test_bedrock_guardrail_disable_exception_on_block_streaming(): async for chunk in result_generator: pass - # Test 2: disable_exception_on_block=True - should NOT raise exception + # Test 2: disable_exception_on_block=True. Streaming can't raise up to the + # endpoint handler (SSE headers already flushed), so the block is delivered + # as a synthetic stream with finish_reason=content_filter and the block + # message as content -- same shape a non-streaming block produces. guardrail_disabled = BedrockGuardrail( guardrailIdentifier="test-guardrail", guardrailVersion="DRAFT", @@ -1526,31 +1529,20 @@ async def test_bedrock_guardrail_disable_exception_on_block_streaming(): ) as mock_post: mock_post.return_value = mock_bedrock_response - # Should NOT raise exception when disable_exception_on_block=True - try: - result_generator = ( - guardrail_disabled.async_post_call_streaming_iterator_hook( - user_api_key_dict=mock_user_api_key_dict, - response=mock_streaming_response(), - request_data=request_data, - ) - ) - - # Consume the generator - should succeed without exceptions - result_chunks = [] - async for chunk in result_generator: - result_chunks.append(chunk) - - # Should have received chunks back even though content was blocked - assert len(result_chunks) > 0 - print( - "✅ Streaming completed without exception when disable_exception_on_block=True" - ) - - except Exception as e: - pytest.fail( - f"Should not raise exception when disable_exception_on_block=True in streaming, but got: {e}" - ) + result_generator = guardrail_disabled.async_post_call_streaming_iterator_hook( + user_api_key_dict=mock_user_api_key_dict, + response=mock_streaming_response(), + request_data=request_data, + ) + chunks = [c async for c in result_generator] + assert chunks, "streaming block should yield synthetic chunks, not empty" + assembled_content = "".join( + (c.choices[0].delta.content or "") + for c in chunks + if getattr(c, "choices", None) and getattr(c.choices[0], "delta", None) + ) + assert assembled_content == "I can't provide that information." + assert chunks[-1].choices[0].finish_reason == "content_filter" @pytest.mark.asyncio diff --git a/tests/litellm_utils_tests/test_cyberark.py b/tests/litellm_utils_tests/test_cyberark.py index 67575d3e781..71daf35a265 100644 --- a/tests/litellm_utils_tests/test_cyberark.py +++ b/tests/litellm_utils_tests/test_cyberark.py @@ -5,6 +5,7 @@ Integration test for CyberArk Conjur Secret Manager. import os import sys import pytest +import yaml from dotenv import load_dotenv load_dotenv() @@ -42,6 +43,82 @@ def create_mock_response(status_code: int, text: str = ""): return mock_response +@pytest.mark.asyncio +async def test_cyberark_write_secret_rejects_yaml_injection(): + """ + Regression test: async_write_secret must reject a secret_name that is not + safe to embed in the Conjur policy body, before any HTTP call is made. + """ + with patch("litellm.proxy.proxy_server.premium_user", True): + malicious_secret_name = "foo\n- !grant\n role: !!admin\n member: attacker" + + mock_sync_client = MagicMock() + mock_async_client = AsyncMock() + + with ( + patch( + "litellm.secret_managers.cyberark_secret_manager._get_httpx_client", + return_value=mock_sync_client, + ), + patch( + "litellm.secret_managers.cyberark_secret_manager.get_async_httpx_client", + return_value=mock_async_client, + ), + ): + cyberark_manager = CyberArkSecretManager() + + response = await cyberark_manager.async_write_secret( + secret_name=malicious_secret_name, + secret_value="sk-1234", + ) + + assert response["status"] == "error" + assert "Invalid secret_name" in response["message"] + # The malicious policy YAML must never reach the wire. + mock_sync_client.client.post.assert_not_called() + mock_async_client.post.assert_not_called() + + +@pytest.mark.parametrize( + "secret_name", + [ + "foo: bar", + "foo # bar", + "plain-alias", + "team/user@example.com", + ], +) +def test_cyberark_ensure_variable_exists_escapes_yaml_metacharacters(secret_name): + """ + Regression test: _ensure_variable_exists must escape secret_name (not just + denylist-check it) so the policy body always parses back to exactly one + '!variable' scalar node holding the untouched secret_name. + """ + with patch("litellm.proxy.proxy_server.premium_user", True): + captured = {} + + def _capture_post(url, headers=None, content=None): + captured["content"] = content + return create_mock_response(status_code=201, text="") + + mock_sync_client = MagicMock() + mock_sync_client.client.post.side_effect = _capture_post + + with patch( + "litellm.secret_managers.cyberark_secret_manager._get_httpx_client", + return_value=mock_sync_client, + ): + cyberark_manager = CyberArkSecretManager() + cyberark_manager._ensure_variable_exists(secret_name) + + policy_yaml = captured["content"] + parsed = yaml.compose(policy_yaml) + assert len(parsed.value) == 1 + node = parsed.value[0] + assert node.tag == "!variable" + assert node.value == secret_name + + @pytest.mark.asyncio async def test_cyberark_write_and_read_secret(): """ diff --git a/tests/litellm_utils_tests/test_hashicorp.py b/tests/litellm_utils_tests/test_hashicorp.py index 3bdf11ea565..9aff7ddc10e 100644 --- a/tests/litellm_utils_tests/test_hashicorp.py +++ b/tests/litellm_utils_tests/test_hashicorp.py @@ -409,6 +409,33 @@ def test_hashicorp_custom_mount_and_prefix(hashicorp_secret_manager): hashicorp_secret_manager.vault_namespace = original_namespace +@pytest.mark.parametrize( + "malicious_secret_name", + [ + "../../../other-app/creds", + "litellm/../../secret", + "foo\nbar", + "foo
bar", + "foo
bar", + "foo\x85bar", + ], +) +def test_hashicorp_get_url_rejects_path_traversal(monkeypatch, malicious_secret_name): + """ + Regression test: get_url must reject an invalid secret_name instead of + building a URL from it. + + Uses monkeypatch + a directly-constructed manager (not the shared + hashicorp_secret_manager fixture) so this runs in CI without real Vault + credentials configured; get_url performs no I/O. + """ + monkeypatch.setenv("HCP_VAULT_TOKEN", "test-token-for-get-url-only") + manager = HashicorpSecretManager() + + with pytest.raises(ValueError): + manager.get_url(malicious_secret_name) + + mock_old_vault_response = { "request_id": "80fafb6a-e96a-4c5b-29fa-ff505ac72201", "lease_id": "", diff --git a/tests/llm_responses_api_testing/base_responses_api.py b/tests/llm_responses_api_testing/base_responses_api.py index 7d2e30f8372..407091a65b3 100644 --- a/tests/llm_responses_api_testing/base_responses_api.py +++ b/tests/llm_responses_api_testing/base_responses_api.py @@ -746,7 +746,8 @@ class BaseResponsesAPITest(ABC): E2E test for Shell tool on OpenAI Responses API. Passes tools=[{"type": "shell", "environment": {"type": "container_auto"}}]; validates that the request is accepted and returns a valid response. - Only runs for OpenAI/Azure (Responses API with shell support). + Only runs for OpenAI; offline coverage for the Azure route lives in + tests/test_litellm/responses/test_responses_api_request_body.py. """ base_completion_call_args = self.get_base_completion_call_args() model = ( @@ -754,8 +755,10 @@ class BaseResponsesAPITest(ABC): or base_completion_call_args.get("model") or "" ) - if "openai/" not in str(model) and "azure/" not in str(model): - pytest.skip("Shell tool e2e is only run for OpenAI/Azure Responses API") + if "openai/" not in str(model): + pytest.skip( + "Shell tool e2e is OpenAI-only; no Azure deployment supports the shell tool yet, re-enable once one exists" + ) tools = [{"type": "shell", "environment": {"type": "container_auto"}}] input_msg = "List files in /mnt/data and show python --version." try: diff --git a/tests/llm_responses_api_testing/test_azure_responses_api.py b/tests/llm_responses_api_testing/test_azure_responses_api.py index fed9e9e11f0..ccef8cbf1e7 100644 --- a/tests/llm_responses_api_testing/test_azure_responses_api.py +++ b/tests/llm_responses_api_testing/test_azure_responses_api.py @@ -2,7 +2,6 @@ import os import sys import pytest import asyncio -from typing import Optional from unittest.mock import patch, AsyncMock sys.path.insert(0, os.path.abspath("../..")) @@ -30,10 +29,6 @@ class TestAzureResponsesAPITest(BaseResponsesAPITest): "api_version": "2025-03-01-preview", } - def get_advanced_model_for_shell_tool(self) -> Optional[str]: - """If specified, overrides the model used by test_responses_api_shell_tool_streaming_sees_shell_output (e.g. openai/gpt-5.2 for shell support).""" - return "azure/gpt-5-mini" - @pytest.mark.asyncio async def test_azure_responses_api_preview_api_version(): diff --git a/tests/proxy_behavior/management/test_team_update.py b/tests/proxy_behavior/management/test_team_update.py index 9cb2b0fecda..23ea89fa74d 100644 --- a/tests/proxy_behavior/management/test_team_update.py +++ b/tests/proxy_behavior/management/test_team_update.py @@ -105,27 +105,35 @@ async def test_team_update_authz_matrix( assert row.team_alias != MARKER_ALIAS, "denied but team mutated" -async def test_team_update_requires_proxy_admin_without_org_context( +async def test_team_update_org_admin_resolved_from_team_without_org_context( proxy_client, prisma, scratch, world ): - """With no organization_id in the body the route gate has no org context - and falls back to proxy-admin-only: an org admin of the team's own org - is 401, PROXY_ADMIN is 200.""" + """With no organization_id in the body the route gate resolves the target + team's org from team_id, so an org admin of the team's own org is allowed + (200), same as PROXY_ADMIN. A team admin of that same team stays denied + (401): the resolution grants org admins access, not team admins.""" await _seed_target(prisma, world, "alpha", scratch.prefix) - denied = await proxy_client.post( + allowed_org_admin = await proxy_client.post( "/team/update", headers={"Authorization": f"Bearer {world.keys[Actor.ORG_ADMIN].cleartext}"}, json={"team_id": scratch.prefix, "team_alias": MARKER_ALIAS}, ) - assert denied.status_code == 401, denied.text + assert allowed_org_admin.status_code == 200, allowed_org_admin.text - allowed = await proxy_client.post( + allowed_proxy_admin = await proxy_client.post( "/team/update", headers={"Authorization": f"Bearer {world.keys[Actor.PROXY_ADMIN].cleartext}"}, json={"team_id": scratch.prefix, "team_alias": MARKER_ALIAS}, ) - assert allowed.status_code == 200, allowed.text + assert allowed_proxy_admin.status_code == 200, allowed_proxy_admin.text + + denied_team_admin = await proxy_client.post( + "/team/update", + headers={"Authorization": f"Bearer {world.keys[Actor.TEAM_ADMIN].cleartext}"}, + json={"team_id": scratch.prefix, "team_alias": MARKER_ALIAS}, + ) + assert denied_team_admin.status_code == 401, denied_team_admin.text # Relocation gate — moving a team to a different org. The scratch team starts diff --git a/tests/test_litellm/enterprise/enterprise_callbacks/send_emails/test_base_email.py b/tests/test_litellm/enterprise/enterprise_callbacks/send_emails/test_base_email.py index c1ccc454305..61303340570 100644 --- a/tests/test_litellm/enterprise/enterprise_callbacks/send_emails/test_base_email.py +++ b/tests/test_litellm/enterprise/enterprise_callbacks/send_emails/test_base_email.py @@ -348,7 +348,9 @@ async def test_get_invitation_link(base_email_logger): result = await base_email_logger._get_invitation_link( user_id="test-user", base_url="http://test.com" ) - assert result == "http://test.com/ui?invitation_id=test-invitation-id" + assert ( + result == "http://test.com/ui/onboarding?invitation_id=test-invitation-id" + ) # Test with None user_id result = await base_email_logger._get_invitation_link( @@ -372,7 +374,7 @@ def test_construct_invitation_link(base_email_logger): result = base_email_logger._construct_invitation_link( invitation_id="test-id-123", base_url="http://test.com" ) - assert result == "http://test.com/ui?invitation_id=test-id-123" + assert result == "http://test.com/ui/onboarding?invitation_id=test-id-123" @pytest.mark.asyncio @@ -408,7 +410,10 @@ async def test_get_invitation_link_creates_new_when_none_exist(base_email_logger assert call_args["user_api_key_dict"].user_id == "test-user" # Verify the returned link uses the new invitation ID - assert result == "http://test.com/ui?invitation_id=new-invitation-id" + assert ( + result + == "http://test.com/ui/onboarding?invitation_id=new-invitation-id" + ) @pytest.mark.asyncio @@ -439,7 +444,10 @@ async def test_get_invitation_link_uses_existing_when_available(base_email_logge mock_create_invitation.assert_not_called() # Verify the returned link uses the existing invitation ID - assert result == "http://test.com/ui?invitation_id=existing-invitation-id" + assert ( + result + == "http://test.com/ui/onboarding?invitation_id=existing-invitation-id" + ) @pytest.mark.asyncio @@ -475,7 +483,10 @@ async def test_get_invitation_link_creates_new_when_list_is_none(base_email_logg assert call_args["user_api_key_dict"].user_id == "test-user" # Verify the returned link uses the new invitation ID - assert result == "http://test.com/ui?invitation_id=new-invitation-from-none" + assert ( + result + == "http://test.com/ui/onboarding?invitation_id=new-invitation-from-none" + ) @pytest.mark.asyncio @@ -495,7 +506,7 @@ async def test_get_email_params_user_invitation( with mock.patch.object( base_email_logger, "_get_invitation_link", - return_value="http://test.com/ui?invitation_id=test-id", + return_value="http://test.com/ui/onboarding?invitation_id=test-id", ): # Test with user invitation event result = await base_email_logger._get_email_params( @@ -509,7 +520,9 @@ async def test_get_email_params_user_invitation( == "https://litellm-listing.s3.amazonaws.com/litellm_logo.png" ) assert result.support_contact == "support@berri.ai" - assert result.base_url == "http://test.com/ui?invitation_id=test-id" + assert ( + result.base_url == "http://test.com/ui/onboarding?invitation_id=test-id" + ) assert result.recipient_email == "test@example.com" diff --git a/tests/test_litellm/expected_responses_api_request/azure_shell_tool.json b/tests/test_litellm/expected_responses_api_request/azure_shell_tool.json new file mode 100644 index 00000000000..b716c518106 --- /dev/null +++ b/tests/test_litellm/expected_responses_api_request/azure_shell_tool.json @@ -0,0 +1,14 @@ +{ + "model": "gpt-5-mini", + "input": "List files in /mnt/data and run python --version.", + "tools": [ + { + "type": "shell", + "environment": { + "type": "container_auto" + } + } + ], + "tool_choice": "auto", + "max_output_tokens": 256 +} diff --git a/tests/test_litellm/integrations/otel/test_otel_v2_components.py b/tests/test_litellm/integrations/otel/test_otel_v2_components.py index 19eef284b91..eb795a64b79 100644 --- a/tests/test_litellm/integrations/otel/test_otel_v2_components.py +++ b/tests/test_litellm/integrations/otel/test_otel_v2_components.py @@ -579,13 +579,10 @@ def _exception_event(span): def test_error_message_recorded_as_full_exception_event_untruncated(): - """Regression for the Elasticsearch keyword/ignore_above:1024 truncation. - - A long error message must survive intact on the standard ``exception`` - event under ``exception.message`` — not get dropped onto a bare string - attribute that backends dynamic-map to a 1024-char ``keyword``. The SDK - must not truncate it either, so a 5000-char message stays 5000 chars. - """ + """The ``exception`` event carries the full untruncated message under + ``exception.message`` so backends that dynamic-map unknown string span + attrs to ``keyword`` (e.g. Elasticsearch with a 1024-char ``ignore_above``) + still see it in full via the semconv-recognized event field.""" from litellm.integrations.otel.model.semconv import Error, ExceptionEvent long_message = "boom: " + "x" * 5000 @@ -596,13 +593,106 @@ def test_error_message_recorded_as_full_exception_event_untruncated(): assert len(event.attributes[ExceptionEvent.MESSAGE]) == len(long_message) > 1024 assert event.attributes[ExceptionEvent.TYPE] == "litellm.APIError" - # error.type stays a low-cardinality attribute; the message does NOT become a - # bare string attribute (which is what got truncated). + # error.type stays a low-cardinality attribute; the exception EVENT field + # ``exception.message`` never becomes a bare string attribute. assert span.attributes[Error.TYPE] == "litellm.APIError" assert ExceptionEvent.MESSAGE not in span.attributes assert span.status.description == long_message +def test_error_details_stamped_as_span_attributes_for_labels_ingest(): + """OTel-defined keys and litellm-specific detail keys both ride span + attributes so backends that flatten attrs into label indexes (Elastic APM + ``labels.*``, Datadog span tags) render them. The exception event with the + full untruncated message stays alongside.""" + from litellm.integrations.otel.model.semconv import Error, ExceptionEvent, LiteLLMError + from litellm.integrations.otel.emitter import SpanEmitter + + cfg = OpenTelemetryV2Config(exporter="in_memory") + provider, exporter = providers.in_memory_provider(cfg) + engine = SpanEmitter(providers.get_tracer(provider, "t"), cfg) + data = LLMCallSpanData( + operation=GenAIOperation.CHAT, + provider="openai", + request_model="gpt-4o", + response_model=None, + response_id=None, + request_params=LLMRequestParams(), + usage=LLMUsage(), + finish_reasons=(), + error=SpanError( + error_type="litellm.BadRequestError", + message="400: violated moderation policy", + code="400", + stack_trace="File proxy_server.py line 8570 ...", + llm_provider="openai", + ), + response_cost=None, + server=None, + identity=RequestIdentity(call_id=None), + ) + engine.emit(SpanRole.LLM_CALL, data) + (span,) = exporter.get_finished_spans() + + # OTel-defined keys (from the ``error.*`` semconv registry). + assert span.attributes[Error.TYPE] == "litellm.BadRequestError" + assert span.attributes[Error.MESSAGE] == "400: violated moderation policy" + # LiteLLM-specific detail keys, under the ``litellm.provider.error.*`` + # vendor namespace, not defined by OTel semconv. + assert span.attributes[LiteLLMError.CODE] == "400" + assert span.attributes[LiteLLMError.STACK_TRACE] == "File proxy_server.py line 8570 ..." + assert span.attributes[LiteLLMError.LLM_PROVIDER] == "openai" + + # The exception event carries the same message on the span too. + event = _exception_event(span) + assert event.attributes[ExceptionEvent.MESSAGE] == "400: violated moderation policy" + + +def test_error_details_omitted_when_span_error_carries_only_message(): + """A guardrail-shape error (message only, no code/traceback/provider) must + not pollute the span with empty-string detail attributes. Only the keys + that carry real data land.""" + from litellm.integrations.otel.model.semconv import Error, LiteLLMError + + span = _emit_error_span("guardrail rejected", error_type="ContentFilter") + + assert span.attributes[Error.TYPE] == "ContentFilter" + assert span.attributes[Error.MESSAGE] == "guardrail rejected" + # LiteLLM-specific detail keys aren't stamped when the SpanError doesn't + # carry them. + assert LiteLLMError.CODE not in span.attributes + assert LiteLLMError.STACK_TRACE not in span.attributes + assert LiteLLMError.LLM_PROVIDER not in span.attributes + + +def test_error_attribute_keys_are_pinned(): + """``error.type`` and ``error.message`` come from the semconv ``error.*`` + registry; the litellm-specific detail keys are vendor keys under + ``litellm.provider.error.*``. Pins the exact strings so the emitted + vocabulary can't drift silently.""" + from litellm.integrations.otel.model.semconv import Error, LiteLLMError + + assert Error.TYPE == "error.type" + assert Error.MESSAGE == "error.message" + assert LiteLLMError.CODE == "litellm.provider.error.code" + assert LiteLLMError.STACK_TRACE == "litellm.provider.error.stack_trace" + assert LiteLLMError.LLM_PROVIDER == "litellm.provider.error.llm_provider" + + +def test_error_message_falls_back_to_error_type_when_message_absent(): + """A ``SpanError(error_type=..., message=None)`` still renders on the span: + the resolved message is the error_type, and it lands on ``error.message``, + the exception event, and the span-status description in lockstep so a + single-source-of-truth view isn't inconsistent.""" + from litellm.integrations.otel.model.semconv import Error, ExceptionEvent + + span = _emit_error_span(message=None, error_type="RateLimitError") + + assert span.attributes[Error.MESSAGE] == "RateLimitError" + assert _exception_event(span).attributes[ExceptionEvent.MESSAGE] == "RateLimitError" + assert span.status.description == "RateLimitError" + + def test_success_span_records_no_exception_event(): from litellm.integrations.otel.emitter import SpanEmitter from litellm.integrations.otel.model.semconv import ExceptionEvent diff --git a/tests/test_litellm/integrations/otel/test_otel_v2_sources_of_truth.py b/tests/test_litellm/integrations/otel/test_otel_v2_sources_of_truth.py index 834a484090f..71be28ea485 100644 --- a/tests/test_litellm/integrations/otel/test_otel_v2_sources_of_truth.py +++ b/tests/test_litellm/integrations/otel/test_otel_v2_sources_of_truth.py @@ -144,11 +144,11 @@ def _all_constants(cls): def test_attribute_keys_are_unique_across_namespaces(): - from litellm.integrations.otel import MCP, Client, JsonRpc, Network + from litellm.integrations.otel import MCP, Client, JsonRpc, LiteLLMError, Network # prefixes are allowed to be substrings; exact keys must not collide. exact = set() - for cls in (GenAI, Error, Server, HTTP, DB, MCP, JsonRpc, Network, Client): + for cls in (GenAI, Error, LiteLLMError, Server, HTTP, DB, MCP, JsonRpc, Network, Client): for key in _all_constants(cls): assert key not in exact, f"duplicate attribute key {key}" exact.add(key) @@ -342,6 +342,47 @@ def test_llm_call_adapter_failure_path(): assert data.error.message == "429 slow down" +def test_llm_call_adapter_carries_error_detail_fields(): + """``_parse_error`` threads the full detail set from ``error_information`` + (``error_code``, ``traceback``, ``llm_provider``) onto ``SpanError`` so the + emitter can stamp them as span attributes.""" + payload = _sample_payload( + status="failure", + error_information={ + "error_class": "BadRequestError", + "error_message": "400 violated moderation policy", + "error_code": "400", + "traceback": "File proxy_server.py line 8570 ...", + "llm_provider": "openai", + }, + ) + data = LLMCallSpanData.from_standard_logging_payload(payload) + assert data.error is not None + assert data.error.error_type == "BadRequestError" + assert data.error.message == "400 violated moderation policy" + assert data.error.code == "400" + assert data.error.stack_trace == "File proxy_server.py line 8570 ..." + assert data.error.llm_provider == "openai" + + +def test_llm_call_adapter_error_details_default_to_none_when_absent(): + """Guardrail-shape payloads carry only ``error_class`` + ``error_message``. + The detail fields must stay ``None`` so the emitter's ``if error.code:`` + guards skip stamping empty attributes.""" + payload = _sample_payload( + status="failure", + error_information={ + "error_class": "ContentFilter", + "error_message": "guardrail rejected", + }, + ) + data = LLMCallSpanData.from_standard_logging_payload(payload) + assert data.error is not None + assert data.error.code is None + assert data.error.stack_trace is None + assert data.error.llm_provider is None + + def test_adapter_is_resilient_to_minimal_payload(): data = LLMCallSpanData.from_standard_logging_payload({}) assert data.request_model == "" diff --git a/tests/test_litellm/integrations/test_custom_guardrail.py b/tests/test_litellm/integrations/test_custom_guardrail.py index 29e9f4529fc..d300f326b9e 100644 --- a/tests/test_litellm/integrations/test_custom_guardrail.py +++ b/tests/test_litellm/integrations/test_custom_guardrail.py @@ -833,6 +833,173 @@ class TestGuardrailSensitiveFieldStripping: assert "sk-secret" not in serialized +class TestGuardrailResponseCredentialMasking: + """LIT-4314 issue B regression: credentials embedded in guardrail_response + (via team callback_vars flowing through data["metadata"]) must be masked at + the construction seam so every downstream sink (SpendLogs, OTel, Langfuse, + custom loggers) sees masked values rather than plaintext. + """ + + def _make_guardrail(self): + from litellm.types.guardrails import GuardrailEventHooks + + return CustomGuardrail( + guardrail_name="test_guardrail", + event_hook=GuardrailEventHooks.pre_call, + ) + + def test_callback_vars_api_key_is_masked(self): + import json + + guardrail = self._make_guardrail() + request_data: dict = {"metadata": {}} + plaintext_key = "lsv2_pt_abcdef1234567890" + + guardrail.add_standard_logging_guardrail_information_to_request_data( + guardrail_json_response={ + "model": "gpt-4o-mini", + "messages": [{"role": "user", "content": "hi"}], + "metadata_snapshot": { + "callback_vars": { + "langsmith_api_key": plaintext_key, + "langsmith_project": "proj-name", + } + }, + }, + request_data=request_data, + guardrail_status="success", + duration=1.0, + ) + + logged = request_data["metadata"]["standard_logging_guardrail_information"][0][ + "guardrail_response" + ] + + masked_key = logged["metadata_snapshot"]["callback_vars"]["langsmith_api_key"] + assert masked_key != plaintext_key + assert "*" in masked_key + assert plaintext_key not in json.dumps(request_data) + + assert logged["model"] == "gpt-4o-mini" + assert logged["messages"] == [{"role": "user", "content": "hi"}] + assert ( + logged["metadata_snapshot"]["callback_vars"]["langsmith_project"] + == "proj-name" + ) + + def test_nested_user_api_key_auth_metadata_is_masked(self): + import json + + guardrail = self._make_guardrail() + request_data: dict = {"metadata": {}} + token_value = "1b01552f6e52e0d41963dd6a185bd6b074624e330999534ca7ff5adfdf622dfc" + + guardrail.add_standard_logging_guardrail_information_to_request_data( + guardrail_json_response={ + "evaluated_metadata": { + "user_api_key_auth": { + "token": token_value, + "api_key": token_value, + "metadata": { + "callback_vars": { + "langsmith_api_key": "lsv2_pt_super_secret_value_1234", + } + }, + } + } + }, + request_data=request_data, + guardrail_status="success", + ) + + serialized = json.dumps(request_data) + assert token_value not in serialized + assert "lsv2_pt_super_secret_value_1234" not in serialized + + def test_secret_fields_pop_still_runs(self): + import json + + guardrail = self._make_guardrail() + request_data: dict = {"metadata": {}} + + guardrail.add_standard_logging_guardrail_information_to_request_data( + guardrail_json_response={ + "model": "gpt-4", + "secret_fields": { + "raw_headers": { + "authorization": "Bearer sk-live-should-not-appear", + } + }, + }, + request_data=request_data, + guardrail_status="success", + ) + + serialized = json.dumps(request_data) + assert "secret_fields" not in serialized + assert "sk-live-should-not-appear" not in serialized + + def test_match_and_regex_redaction_still_runs(self): + guardrail = self._make_guardrail() + request_data: dict = {"metadata": {}} + + guardrail.add_standard_logging_guardrail_information_to_request_data( + guardrail_json_response={ + "filters": [{"regex": r"\d{3}-\d{2}-\d{4}", "action": "BLOCKED"}] + }, + request_data=request_data, + guardrail_status="success", + ) + + slg = request_data["metadata"]["standard_logging_guardrail_information"][0] + assert slg["guardrail_response"]["filters"][0]["regex"] == "[REDACTED]" + + def test_scalar_types_pass_through_unchanged(self): + guardrail = self._make_guardrail() + request_data: dict = {"metadata": {}} + + guardrail.add_standard_logging_guardrail_information_to_request_data( + guardrail_json_response={ + "flagged": True, + "score": 0.94, + "tokens_used": 42, + "categories": ["pii", "toxicity"], + }, + request_data=request_data, + guardrail_status="success", + ) + + logged = request_data["metadata"]["standard_logging_guardrail_information"][0][ + "guardrail_response" + ] + assert logged["flagged"] is True + assert logged["score"] == 0.94 + assert logged["tokens_used"] == 42 + assert logged["categories"] == ["pii", "toxicity"] + + def test_masking_reveals_prefix_and_suffix(self): + guardrail = self._make_guardrail() + request_data: dict = {"metadata": {}} + plaintext = "lsv2_pt_abcdef1234567890" + + guardrail.add_standard_logging_guardrail_information_to_request_data( + guardrail_json_response={ + "metadata_snapshot": { + "callback_vars": {"langsmith_api_key": plaintext} + } + }, + request_data=request_data, + guardrail_status="success", + ) + + masked = request_data["metadata"]["standard_logging_guardrail_information"][0][ + "guardrail_response" + ]["metadata_snapshot"]["callback_vars"]["langsmith_api_key"] + assert masked != plaintext + assert masked.startswith(plaintext[:4]) + assert masked.endswith(plaintext[-4:]) + + class TestCustomGuardrailPassthroughSupport: """Tests for passthrough endpoint guardrail support - Issue fixes.""" diff --git a/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py b/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py index 0e6c6061b46..ca23e61352e 100644 --- a/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py +++ b/tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py @@ -507,6 +507,104 @@ def test_generic_cost_per_token_gpt55_pro(): ) +@pytest.mark.parametrize( + "model,input_cost,output_cost,cache_read_cost,cache_write_cost", + [ + ("gpt-5.6", 5e-6, 3e-5, 5e-7, 6.25e-6), + ("gpt-5.6-sol", 5e-6, 3e-5, 5e-7, 6.25e-6), + ("gpt-5.6-terra", 2.5e-6, 1.5e-5, 2.5e-7, 3.125e-6), + ("gpt-5.6-luna", 1e-6, 6e-6, 1e-7, 1.25e-6), + ], +) +def test_generic_cost_per_token_gpt56( + model, input_cost, output_cost, cache_read_cost, cache_write_cost +): + """gpt-5.6 (sol/terra/luna): base pricing + new cache-write cost. + + Cache writes are billed at 1.25x the uncached input rate for this family. + """ + custom_llm_provider = "openai" + os.environ["LITELLM_LOCAL_MODEL_COST_MAP"] = "True" + litellm.model_cost = litellm.get_model_cost_map(url="") + + model_cost_map = litellm.model_cost[model] + + assert model_cost_map["input_cost_per_token"] == input_cost + assert model_cost_map["output_cost_per_token"] == output_cost + assert model_cost_map["cache_read_input_token_cost"] == cache_read_cost + assert model_cost_map["cache_creation_input_token_cost"] == cache_write_cost + assert model_cost_map["litellm_provider"] == "openai" + assert model_cost_map["mode"] == "chat" + assert model_cost_map["cache_creation_input_token_cost"] == pytest.approx( + input_cost * 1.25 + ) + assert model_cost_map["max_input_tokens"] == 1050000 + assert model_cost_map["input_cost_per_token_above_272k_tokens"] == pytest.approx( + input_cost * 2 + ) + assert model_cost_map["output_cost_per_token_above_272k_tokens"] == pytest.approx( + output_cost * 1.5 + ) + + prompt_tokens = 1000 + completion_tokens = 500 + usage = Usage( + prompt_tokens=prompt_tokens, + completion_tokens=completion_tokens, + total_tokens=prompt_tokens + completion_tokens, + ) + prompt_cost, completion_cost = generic_cost_per_token( + model=model, + usage=usage, + custom_llm_provider=custom_llm_provider, + ) + assert round(prompt_cost, 10) == round(input_cost * prompt_tokens, 10) + assert round(completion_cost, 10) == round(output_cost * completion_tokens, 10) + + +@pytest.mark.parametrize( + "model,input_cost,output_cost,cache_read_cost", + [ + ("azure/gpt-5.6", 5e-6, 3e-5, 5e-7), + ("azure/gpt-5.6-sol", 5e-6, 3e-5, 5e-7), + ("azure/gpt-5.6-terra", 2.5e-6, 1.5e-5, 2.5e-7), + ("azure/gpt-5.6-luna", 1e-6, 6e-6, 1e-7), + ("azure/us/gpt-5.6", 5.5e-6, 3.3e-5, 5.5e-7), + ("azure/eu/gpt-5.6-terra", 2.75e-6, 1.65e-5, 2.75e-7), + ("azure/eu/gpt-5.6-luna", 1.1e-6, 6.6e-6, 1.1e-7), + ], +) +def test_generic_cost_per_token_azure_gpt56( + model, input_cost, output_cost, cache_read_cost +): + """Azure gpt-5.6 (global + us/eu regional): pricing mirrors the openai + family for global deployments and carries the standard 10% regional uplift. + """ + os.environ["LITELLM_LOCAL_MODEL_COST_MAP"] = "True" + litellm.model_cost = litellm.get_model_cost_map(url="") + + model_cost_map = litellm.model_cost[model] + assert model_cost_map["litellm_provider"] == "azure" + assert model_cost_map["input_cost_per_token"] == input_cost + assert model_cost_map["output_cost_per_token"] == output_cost + assert model_cost_map["cache_read_input_token_cost"] == cache_read_cost + + prompt_tokens = 1000 + completion_tokens = 500 + usage = Usage( + prompt_tokens=prompt_tokens, + completion_tokens=completion_tokens, + total_tokens=prompt_tokens + completion_tokens, + ) + prompt_cost, completion_cost = generic_cost_per_token( + model=model, + usage=usage, + custom_llm_provider="azure", + ) + assert round(prompt_cost, 10) == round(input_cost * prompt_tokens, 10) + assert round(completion_cost, 10) == round(output_cost * completion_tokens, 10) + + @pytest.mark.parametrize( "model,expected_none,expected_xhigh,expected_minimal", [ diff --git a/tests/test_litellm/litellm_core_utils/prompt_templates/test_litellm_core_utils_prompt_templates_factory.py b/tests/test_litellm/litellm_core_utils/prompt_templates/test_litellm_core_utils_prompt_templates_factory.py index 1d5289737f1..bcda88ea609 100644 --- a/tests/test_litellm/litellm_core_utils/prompt_templates/test_litellm_core_utils_prompt_templates_factory.py +++ b/tests/test_litellm/litellm_core_utils/prompt_templates/test_litellm_core_utils_prompt_templates_factory.py @@ -3085,3 +3085,84 @@ def test_bedrock_converse_messages_pt_document_rejects_url_source(): _bedrock_converse_messages_pt( messages, "anthropic.claude-sonnet-4-6", "bedrock" ) + + +def _collect_cache_points(blocks): + return [ + block["cachePoint"] + for message in blocks + for block in message["content"] + if "cachePoint" in block + ] + + +@pytest.mark.parametrize( + "messages", + [ + [ + { + "role": "user", + "content": [ + { + "type": "text", + "text": "conversation history", + "cache_control": {"type": "ephemeral", "ttl": "1h"}, + } + ], + }, + ], + [ + {"role": "user", "content": "hello"}, + { + "role": "assistant", + "content": [ + { + "type": "text", + "text": "assistant reply", + "cache_control": {"type": "ephemeral", "ttl": "1h"}, + } + ], + }, + ], + ], +) +def test_bedrock_converse_message_level_cache_point_preserves_ttl(messages): + """ + Regression for https://github.com/BerriAI/litellm/issues/32154: message-level + cache_control ttl was silently dropped because the message-level + _get_cache_point_block call sites never passed model=, so multi-turn prefixes + fell back to the 5m default while the system prompt kept 1h, churning the + cache every turn on models like Opus 4.8. + """ + result = _bedrock_converse_messages_pt( + messages=messages, + model="eu.anthropic.claude-opus-4-8", + llm_provider="bedrock", + ) + + cache_points = _collect_cache_points(result) + assert cache_points == [{"type": "default", "ttl": "1h"}] + + +@pytest.mark.asyncio +async def test_bedrock_converse_message_level_cache_point_preserves_ttl_async(): + messages = [ + { + "role": "user", + "content": [ + { + "type": "text", + "text": "conversation history", + "cache_control": {"type": "ephemeral", "ttl": "1h"}, + } + ], + }, + ] + + result = await BedrockConverseMessagesProcessor._bedrock_converse_messages_pt_async( + messages=messages, + model="eu.anthropic.claude-opus-4-8", + llm_provider="bedrock", + ) + + assert _collect_cache_points(result) == [{"type": "default", "ttl": "1h"}] diff --git a/tests/test_litellm/litellm_core_utils/test_health_check_helpers.py b/tests/test_litellm/litellm_core_utils/test_health_check_helpers.py index 02d72c89e80..f0d91224614 100644 --- a/tests/test_litellm/litellm_core_utils/test_health_check_helpers.py +++ b/tests/test_litellm/litellm_core_utils/test_health_check_helpers.py @@ -14,6 +14,7 @@ from litellm.constants import LITTELM_INTERNAL_HEALTH_SERVICE_ACCOUNT_NAME from litellm.litellm_core_utils.health_check_helpers import HealthCheckHelpers from litellm.main import ahealth_check from litellm.proxy._types import UserAPIKeyAuth +from litellm.types.utils import LIST_BATCHES_SUPPORTED_PROVIDERS def test_update_model_params_with_health_check_tracking_information(): @@ -140,3 +141,206 @@ async def test_ahealth_check_failure_masks_raw_request_headers(): assert headers["Content-Type"] == "application/json" print(f"Masked Authorization header: {headers.get('Authorization', 'NOT FOUND')}") + + +@pytest.mark.asyncio +async def test_batch_health_check_bridges_metadata_into_logging_obj(): + """_batch_health_check must call update_from_kwargs on the pre-injected + logging object so callbacks receive identity/tracking fields in + model_call_details["litellm_params"]["metadata"].""" + mock_logging_obj = MagicMock() + mock_logging_obj.update_from_kwargs = MagicMock() + + litellm_metadata = { + "tags": [LITTELM_INTERNAL_HEALTH_SERVICE_ACCOUNT_NAME], + "user_api_key_alias": "health-check-key", + } + + filtered_model_params = { + "model": "openai/gpt-4", + "api_base": "https://api.openai.com", + "litellm_logging_obj": mock_logging_obj, + "litellm_metadata": litellm_metadata, + } + + with patch("litellm.alist_batches", new_callable=AsyncMock, return_value={}): + await HealthCheckHelpers._batch_health_check( + custom_llm_provider="openai", + model_params={"model": "openai/gpt-4"}, + filtered_model_params=filtered_model_params, + ) + + mock_logging_obj.update_from_kwargs.assert_called_once() + call_kwargs = mock_logging_obj.update_from_kwargs.call_args[1] + assert call_kwargs["model"] == "openai/gpt-4" + assert call_kwargs["kwargs"] is filtered_model_params + assert call_kwargs["litellm_params"] == {"api_base": "https://api.openai.com"} + + +@pytest.mark.asyncio +async def test_batch_health_check_omits_api_base_when_absent(): + """api_base must not appear in litellm_params when the provider resolves + it implicitly (bedrock, vertex, gemini).""" + mock_logging_obj = MagicMock() + mock_logging_obj.update_from_kwargs = MagicMock() + + litellm_metadata = {"tags": [LITTELM_INTERNAL_HEALTH_SERVICE_ACCOUNT_NAME]} + + filtered_model_params = { + "model": "bedrock/anthropic.claude-v2", + "litellm_logging_obj": mock_logging_obj, + "litellm_metadata": litellm_metadata, + } + + with patch("litellm.acompletion", new_callable=AsyncMock, return_value={}): + await HealthCheckHelpers._batch_health_check( + custom_llm_provider="bedrock", + model_params={"model": "bedrock/anthropic.claude-v2"}, + filtered_model_params=filtered_model_params, + ) + + call_kwargs = mock_logging_obj.update_from_kwargs.call_args[1] + assert call_kwargs["litellm_params"] is None + + +@pytest.mark.asyncio +async def test_batch_health_check_skips_bridge_when_no_logging_obj(): + """When litellm_logging_obj is absent, dispatch still proceeds.""" + litellm_metadata = {"tags": [LITTELM_INTERNAL_HEALTH_SERVICE_ACCOUNT_NAME]} + + filtered_model_params = { + "model": "openai/gpt-4", + "litellm_metadata": litellm_metadata, + } + + with patch( + "litellm.alist_batches", new_callable=AsyncMock, return_value={} + ) as mock_alist: + await HealthCheckHelpers._batch_health_check( + custom_llm_provider="openai", + model_params={"model": "openai/gpt-4"}, + filtered_model_params=filtered_model_params, + ) + mock_alist.assert_called_once() + + +@pytest.mark.asyncio +async def test_batch_health_check_uses_alist_batches_for_supported_providers(): + """Providers in LIST_BATCHES_SUPPORTED_PROVIDERS dispatch to alist_batches.""" + mock_logging_obj = MagicMock() + mock_logging_obj.update_from_kwargs = MagicMock() + + litellm_metadata = {"tags": [LITTELM_INTERNAL_HEALTH_SERVICE_ACCOUNT_NAME]} + + for provider in LIST_BATCHES_SUPPORTED_PROVIDERS: + filtered_model_params = { + "model": f"{provider}/some-model", + "litellm_logging_obj": mock_logging_obj, + "litellm_metadata": litellm_metadata, + } + + with patch( + "litellm.alist_batches", new_callable=AsyncMock, return_value={} + ) as mock_alist: + await HealthCheckHelpers._batch_health_check( + custom_llm_provider=provider, + model_params={"model": f"{provider}/some-model"}, + filtered_model_params=filtered_model_params, + ) + mock_alist.assert_called_once() + + +@pytest.mark.asyncio +async def test_batch_health_check_falls_back_to_acompletion_for_unsupported(): + """Providers not in LIST_BATCHES_SUPPORTED_PROVIDERS fall back to acompletion.""" + mock_logging_obj = MagicMock() + mock_logging_obj.update_from_kwargs = MagicMock() + + litellm_metadata = {"tags": [LITTELM_INTERNAL_HEALTH_SERVICE_ACCOUNT_NAME]} + + filtered_model_params = { + "model": "bedrock/anthropic.claude-v2", + "litellm_logging_obj": mock_logging_obj, + "litellm_metadata": litellm_metadata, + } + + model_params = {"model": "bedrock/anthropic.claude-v2", "messages": []} + + with ( + patch("litellm.alist_batches", new_callable=AsyncMock) as mock_alist, + patch("litellm.acompletion", new_callable=AsyncMock, return_value={}) as mock_acompletion, + ): + await HealthCheckHelpers._batch_health_check( + custom_llm_provider="bedrock", + model_params=model_params, + filtered_model_params=filtered_model_params, + ) + mock_alist.assert_not_called() + mock_acompletion.assert_called_once_with(**model_params) + + +class _FakeWebsocketConnect: + def __init__(self, calls, url, **kwargs): + calls.append({"url": url, **kwargs}) + + async def __aenter__(self): + return self + + async def __aexit__(self, exc_type, exc, tb): + return False + + +@pytest.mark.asyncio +async def test_realtime_health_check_uses_model_level_vertex_params(): + """Regression test: realtime health checks must resolve vertex_credentials, + vertex_project, and vertex_location from the model row's params instead of + falling back to process-global VERTEXAI_* settings.""" + import litellm + from litellm.realtime_api import main as realtime_main + + fake_vertex_base = MagicMock() + fake_vertex_base.get_vertex_region = MagicMock(return_value="us-central1") + fake_vertex_base._ensure_access_token_async = AsyncMock( + return_value=("model-level-token", "model-level-project") + ) + connect_calls = [] + + with ( + patch.object(realtime_main, "vertex_llm_base", fake_vertex_base), + patch( + "websockets.connect", + lambda url, **kwargs: _FakeWebsocketConnect(connect_calls, url, **kwargs), + ), + patch.object( + HealthCheckHelpers, + "_update_model_params_with_health_check_tracking_information", + staticmethod(lambda model_params: model_params), + ), + ): + result = await litellm.ahealth_check( + model_params={ + "model": "vertex_ai/gemini-live-2.5-flash-native-audio", + "vertex_credentials": '{"type":"service_account"}', + "vertex_project": "model-level-project", + "vertex_location": "us-central1", + }, + mode="realtime", + ) + + assert result == {} + fake_vertex_base.get_vertex_region.assert_called_once_with( + vertex_region="us-central1", model="gemini-live-2.5-flash-native-audio" + ) + fake_vertex_base._ensure_access_token_async.assert_called_once_with( + credentials='{"type":"service_account"}', + project_id="model-level-project", + custom_llm_provider="vertex_ai", + ) + assert connect_calls[0]["url"] == ( + "wss://us-central1-aiplatform.googleapis.com/ws/" + "google.cloud.aiplatform.v1.LlmBidiService/BidiGenerateContent" + ) + assert connect_calls[0]["additional_headers"] == { + "Authorization": "Bearer model-level-token", + "x-goog-user-project": "model-level-project", + } diff --git a/tests/test_litellm/litellm_core_utils/test_sensitive_data_masker.py b/tests/test_litellm/litellm_core_utils/test_sensitive_data_masker.py index 9b0116e6979..ba8540f81e3 100644 --- a/tests/test_litellm/litellm_core_utils/test_sensitive_data_masker.py +++ b/tests/test_litellm/litellm_core_utils/test_sensitive_data_masker.py @@ -240,3 +240,78 @@ def test_mask_sensitive_structure_masks_credentials_nested_in_config_shape(): [{"primary-group": [{"model": "gpt-4o", "api_key": secret}]}] ) assert secret not in str(masked) + + +def test_mask_credentials_in_payload_preserves_none_and_scalars(): + """The payload variant does not distort JSON-shaped values: None stays None, + ints/floats/bools stay themselves, lists stay lists. This is what makes it + safe for logging pipelines that persist the record verbatim.""" + from litellm.litellm_core_utils.sensitive_data_masker import mask_credentials_in_payload + + result = mask_credentials_in_payload( + { + "reason": None, + "confidence": 0.42, + "flagged": True, + "tokens_used": 17, + "categories": ["pii", "toxicity"], + "nested": {"end_user_id": None}, + } + ) + assert result == { + "reason": None, + "confidence": 0.42, + "flagged": True, + "tokens_used": 17, + "categories": ["pii", "toxicity"], + "nested": {"end_user_id": None}, + } + + +def test_mask_credentials_in_payload_masks_inside_pydantic_models(): + """A Pydantic model reached during the walk gets dumped to a dict so its + sensitive-named string fields are masked. Without this the credentials + inside a nested ``UserAPIKeyAuth`` in a guardrail_response reach the + logging pipeline unmasked once JSON serialization flattens it.""" + from pydantic import BaseModel + + from litellm.litellm_core_utils.sensitive_data_masker import mask_credentials_in_payload + + class Auth(BaseModel): + token: str = "1b01552f6e52e0d41963dd6a185bd6b074624e330999534ca7ff5adfdf622dfc" + team_alias: str = "acme" + + result = mask_credentials_in_payload({"user_api_key_auth": Auth()}) + auth_dict = result["user_api_key_auth"] + assert isinstance(auth_dict, dict) + assert auth_dict["team_alias"] == "acme" + assert ( + auth_dict["token"] + != "1b01552f6e52e0d41963dd6a185bd6b074624e330999534ca7ff5adfdf622dfc" + ) + assert "*" in auth_dict["token"] + + +def test_mask_credentials_in_payload_masks_only_sensitive_string_leaves(): + """Sensitive-named string leaves get masked; sibling non-string values + (including None) under the same key stay verbatim.""" + from litellm.litellm_core_utils.sensitive_data_masker import mask_credentials_in_payload + + plaintext = "lsv2_pt_abcdef1234567890" + result = mask_credentials_in_payload( + { + "model": "gpt-4o-mini", + "callback_vars": { + "langsmith_api_key": plaintext, + "langsmith_project": "proj", + "extra_token_count": 5, + }, + } + ) + assert result["model"] == "gpt-4o-mini" + assert result["callback_vars"]["langsmith_project"] == "proj" + assert result["callback_vars"]["extra_token_count"] == 5 + masked = result["callback_vars"]["langsmith_api_key"] + assert masked != plaintext + assert masked.startswith(plaintext[:4]) + assert masked.endswith(plaintext[-4:]) diff --git a/tests/test_litellm/llms/bedrock/chat/test_converse_transformation.py b/tests/test_litellm/llms/bedrock/chat/test_converse_transformation.py index fe060e2e40d..9f2a4168dec 100644 --- a/tests/test_litellm/llms/bedrock/chat/test_converse_transformation.py +++ b/tests/test_litellm/llms/bedrock/chat/test_converse_transformation.py @@ -5671,3 +5671,99 @@ async def test_grounding_source_and_query_rendered_as_text(): user_content = result[0]["content"] assert {"text": "Tokyo is the capital of Japan."} in user_content assert {"text": "What is the capital of Japan?"} in user_content + + +def _agentic_messages_with_ttl(ttl_target: str): + """A tool-loop conversation with `ttl: 1h` cache_control at `ttl_target`: + 'user', 'tool_call' (per-tool-call, on the assistant's tool call), or + 'tool' (message-level, on the tool result - where + `cache_control_injection_points` with `index: -1` lands mid-loop). + + Message-level cache_control on a content-less assistant message emits no + cachePoint at all today (a separate gap, orthogonal to ttl); per-tool-call + placement covers that message, so it's excluded from the params below.""" + user: dict = {"role": "user", "content": "optimize this kernel " * 60} + assistant: dict = { + "role": "assistant", + "content": None, + "tool_calls": [ + { + "id": "call_1", + "type": "function", + "function": {"name": "evaluate", "arguments": "{}"}, + } + ], + } + tool: dict = {"role": "tool", "tool_call_id": "call_1", "content": "score: 42"} + ttl_cc = {"type": "ephemeral", "ttl": "1h"} + if ttl_target == "user": + user["cache_control"] = ttl_cc + elif ttl_target == "tool_call": + assistant["tool_calls"][0]["cache_control"] = ttl_cc + elif ttl_target == "tool": + tool["cache_control"] = ttl_cc + return [user, assistant, tool] + + +def _collect_cache_points(result): + return [ + block["cachePoint"] + for message in result + for block in message.get("content") or [] + if "cachePoint" in block + ] + + +@pytest.mark.parametrize("ttl_target", ["user", "tool_call", "tool"]) +@pytest.mark.asyncio +async def test_message_level_cache_control_honors_ttl_for_supported_model( + ttl_target, +): + """Message- and tool-call-level cache_control must carry `ttl` onto the + emitted cachePoint for models that support extended caching, mirroring the + system-message path. Regression test for the gap left by the system-only + fix: the message paths called `_get_cache_point_block` without `model` (or + hardcoded `{"type": "default"}`), silently downgrading 1h to 5m.""" + from litellm.litellm_core_utils.prompt_templates.factory import ( + BedrockConverseMessagesProcessor, + _bedrock_converse_messages_pt, + ) + + messages = _agentic_messages_with_ttl(ttl_target) + + result = _bedrock_converse_messages_pt( + messages=messages, + model="global.anthropic.claude-opus-4-7", + llm_provider="bedrock_converse", + ) + async_result = ( + await BedrockConverseMessagesProcessor._bedrock_converse_messages_pt_async( + messages=messages, + model="global.anthropic.claude-opus-4-7", + llm_provider="bedrock_converse", + ) + ) + assert result == async_result + + cache_points = _collect_cache_points(result) + assert len(cache_points) == 1 + assert cache_points[0].get("ttl") == "1h" + + +@pytest.mark.parametrize("ttl_target", ["user", "tool_call", "tool"]) +def test_message_level_cache_control_drops_ttl_for_unsupported_model(ttl_target): + """Models outside the extended-caching allow-list must keep emitting the + plain `{"type": "default"}` cachePoint (Bedrock rejects `ttl` for them).""" + from litellm.litellm_core_utils.prompt_templates.factory import ( + _bedrock_converse_messages_pt, + ) + + result = _bedrock_converse_messages_pt( + messages=_agentic_messages_with_ttl(ttl_target), + model="anthropic.claude-3-5-sonnet-20240620-v1:0", + llm_provider="bedrock_converse", + ) + + cache_points = _collect_cache_points(result) + assert len(cache_points) == 1 + assert "ttl" not in cache_points[0] diff --git a/tests/test_litellm/llms/bedrock/messages/invoke_transformations/test_anthropic_claude3_transformation.py b/tests/test_litellm/llms/bedrock/messages/invoke_transformations/test_anthropic_claude3_transformation.py index d7a62aae38b..532c6ff3598 100644 --- a/tests/test_litellm/llms/bedrock/messages/invoke_transformations/test_anthropic_claude3_transformation.py +++ b/tests/test_litellm/llms/bedrock/messages/invoke_transformations/test_anthropic_claude3_transformation.py @@ -2090,3 +2090,174 @@ def test_bedrock_clear_thinking_leaves_enabled_thinking_on_non_adaptive_model(): assert changed is False assert request["thinking"] == {"type": "enabled", "budget_tokens": 8000} assert "output_config" not in request + + +@pytest.fixture +def local_beta_headers_config(monkeypatch): + from litellm.anthropic_beta_headers_manager import reload_beta_headers_config + + monkeypatch.setenv("LITELLM_LOCAL_ANTHROPIC_BETA_HEADERS", "True") + reload_beta_headers_config() + yield + monkeypatch.delenv("LITELLM_LOCAL_ANTHROPIC_BETA_HEADERS", raising=False) + reload_beta_headers_config() + + +def test_bedrock_messages_preserves_clear_tool_uses_context_management_and_adds_beta( + local_beta_headers_config, +): + """ + LIT-3393: Bedrock InvokeModel supports automatic tool-call clearing via + ``clear_tool_uses_20250919`` under the ``context-management-2025-06-27`` + beta. Before the LIT-3393 fix, the transformation stripped this edit (only + ``compact_20260112`` survived) AND the beta was filtered out by + ``filter_and_transform_beta_headers`` for ``bedrock``, producing a Bedrock + 400 ``"context_management: Extra inputs are not permitted"``. + + Post-fix, the edit must reach the body and the beta must reach + ``anthropic_beta``. + + AWS docs ("Automatic tool call clearing (Beta)"): + https://docs.aws.amazon.com/bedrock/latest/userguide/model-parameters-anthropic-claude-messages-tool-use.md + """ + from litellm.types.router import GenericLiteLLMParams + + cfg = AmazonAnthropicClaudeMessagesConfig() + messages = [{"role": "user", "content": [{"type": "text", "text": "Hi"}]}] + optional_params = { + "max_tokens": 4096, + "context_management": { + "edits": [{"type": "clear_tool_uses_20250919"}] + }, + } + + result = cfg.transform_anthropic_messages_request( + model="anthropic.claude-haiku-4-5-20251001-v1:0", + messages=messages, + anthropic_messages_optional_request_params=optional_params, + litellm_params=GenericLiteLLMParams(), + headers={}, + ) + + assert result.get("context_management") == { + "edits": [{"type": "clear_tool_uses_20250919"}] + }, "clear_tool_uses_20250919 edit must reach Bedrock InvokeModel body" + assert "context-management-2025-06-27" in result.get("anthropic_beta", []), ( + "context-management-2025-06-27 beta must reach the InvokeModel body so " + "the tool-call-clearing edit is accepted" + ) + + +def test_bedrock_messages_preserves_mixed_compact_and_clear_tool_uses_edits( + local_beta_headers_config, +): + """ + LIT-3393: a request mixing ``compact_20260112`` and + ``clear_tool_uses_20250919`` must keep BOTH edits and emit BOTH + anthropic-beta values (``compact-2026-01-12`` + ``context-management-2025-06-27``). + """ + from litellm.types.router import GenericLiteLLMParams + + cfg = AmazonAnthropicClaudeMessagesConfig() + messages = [{"role": "user", "content": [{"type": "text", "text": "Hi"}]}] + optional_params = { + "max_tokens": 4096, + "context_management": { + "edits": [ + {"type": "compact_20260112"}, + {"type": "clear_tool_uses_20250919"}, + ] + }, + } + + result = cfg.transform_anthropic_messages_request( + model="anthropic.claude-sonnet-4-6-20250929-v1:0", + messages=messages, + anthropic_messages_optional_request_params=optional_params, + litellm_params=GenericLiteLLMParams(), + headers={}, + ) + + cm = result.get("context_management") + assert cm is not None + edit_types = sorted(e.get("type") for e in cm["edits"]) + assert edit_types == ["clear_tool_uses_20250919", "compact_20260112"] + + betas = result.get("anthropic_beta", []) + assert "compact-2026-01-12" in betas + assert "context-management-2025-06-27" in betas + + +def test_bedrock_messages_filters_clear_thinking_keeps_clear_tool_uses( + local_beta_headers_config, +): + """ + LIT-3393: ``clear_thinking_20251015`` remains LiteLLM-internal (consumed via + thinking-injection) and MUST be stripped from the body, while + ``clear_tool_uses_20250919`` (officially supported on Bedrock InvokeModel) + survives in the same request. + """ + from litellm.types.router import GenericLiteLLMParams + + cfg = AmazonAnthropicClaudeMessagesConfig() + messages = [{"role": "user", "content": [{"type": "text", "text": "Hi"}]}] + optional_params = { + "max_tokens": 4096, + "context_management": { + "edits": [ + {"type": "clear_thinking_20251015", "keep": "all"}, + {"type": "clear_tool_uses_20250919"}, + ] + }, + } + + result = cfg.transform_anthropic_messages_request( + model="anthropic.claude-opus-4-7", + messages=messages, + anthropic_messages_optional_request_params=optional_params, + litellm_params=GenericLiteLLMParams(), + headers={}, + ) + + cm = result.get("context_management") + assert cm is not None + assert [e.get("type") for e in cm["edits"]] == [ + "clear_tool_uses_20250919" + ], "clear_thinking_20251015 must still be stripped (LiteLLM-internal)" + + betas = result.get("anthropic_beta", []) + assert "context-management-2025-06-27" in betas + # ``compact-2026-01-12`` was not requested. + assert "compact-2026-01-12" not in betas + + +def test_filter_and_transform_beta_headers_passes_context_management_for_bedrock( + local_beta_headers_config, +): + """ + LIT-3393: ``anthropic_beta_headers_config.json`` previously mapped + ``bedrock.context-management-2025-06-27`` to ``null``, so + ``filter_and_transform_beta_headers`` dropped the header even when the + transformation tried to set it. This regression guard locks the bundled + mapping in place. + + Pinned to the bundled local config via ``LITELLM_LOCAL_ANTHROPIC_BETA_HEADERS`` + so the assertion is not subject to whatever the upstream remote currently + serves or what previous tests left in the module cache. + """ + from litellm.anthropic_beta_headers_manager import filter_and_transform_beta_headers + + out = filter_and_transform_beta_headers( + ["context-management-2025-06-27"], + provider="bedrock", + ) + assert out == ["context-management-2025-06-27"] + + # Bedrock_converse genuinely lacks it per AWS docs; this guard prevents + # an accidental flip there. + out_converse = filter_and_transform_beta_headers( + ["context-management-2025-06-27"], + provider="bedrock_converse", + ) + assert out_converse == [] + diff --git a/tests/test_litellm/llms/openai/test_is_model_gpt_5_model.py b/tests/test_litellm/llms/openai/test_is_model_gpt_5_model.py index 02dd9dade0a..1095819c98c 100644 --- a/tests/test_litellm/llms/openai/test_is_model_gpt_5_model.py +++ b/tests/test_litellm/llms/openai/test_is_model_gpt_5_model.py @@ -50,6 +50,10 @@ GPT5_MODELS = [ "gpt-5.5-pro", "gpt-5.5-2026-04-23", # dated variant "gpt-5.5-pro-2026-04-23", # dated variant + "gpt-5.6", + "gpt-5.6-sol", + "gpt-5.6-terra", + "gpt-5.6-luna", "gpt-5.1-chat", # versioned chat — THE KEY REGRESSION CASE "gpt-5.2-chat", # versioned chat — also a regression case "gpt-5.3-chat", # versioned chat — THE KEY REGRESSION CASE @@ -112,6 +116,45 @@ class TestOpenAIGPT5ConfigIsModelGpt5Model: ), f"Expected '{model}' (gpt-5-chat family) NOT to be on the GPT-5 path" +# Models that are gpt-5.4 or newer. main.py gates the automatic switch to the +# /v1/responses bridge (when reasoning_effort is set and tools are passed) on +# is_model_gpt_5_4_plus_model, so the gpt-5.6 family must land on the True side. +GPT5_4_PLUS_MODELS = [ + "gpt-5.4", + "gpt-5.5", + "gpt-5.5-pro", + "gpt-5.6", + "gpt-5.6-sol", + "gpt-5.6-terra", + "gpt-5.6-luna", + "openai/gpt-5.6-sol", +] + +GPT5_PRE_5_4_MODELS = [ + "gpt-5", + "gpt-5.1", + "gpt-5.2", + "gpt-5.3", + "gpt-5.3-chat", + "gpt-4o", +] + + +class TestOpenAIGPT5ConfigIsModelGpt54PlusModel: + + @pytest.mark.parametrize("model", GPT5_4_PLUS_MODELS) + def test_gpt5_4_plus_models_are_classified_as_5_4_plus(self, model: str): + assert OpenAIGPT5Config.is_model_gpt_5_4_plus_model( + model + ), f"Expected '{model}' to be classified as gpt-5.4-or-newer" + + @pytest.mark.parametrize("model", GPT5_PRE_5_4_MODELS) + def test_pre_5_4_models_are_not_classified_as_5_4_plus(self, model: str): + assert not OpenAIGPT5Config.is_model_gpt_5_4_plus_model( + model + ), f"Expected '{model}' NOT to be classified as gpt-5.4-or-newer" + + # --------------------------------------------------------------------------- # AzureOpenAIGPT5Config # --------------------------------------------------------------------------- diff --git a/tests/test_litellm/llms/openai_like/test_meta_provider.py b/tests/test_litellm/llms/openai_like/test_meta_provider.py new file mode 100644 index 00000000000..11b78828da6 --- /dev/null +++ b/tests/test_litellm/llms/openai_like/test_meta_provider.py @@ -0,0 +1,224 @@ +""" +Tests for the Meta Model API (Muse Spark) provider configuration and integration. +""" + +import litellm + + +class TestMetaProviderConfig: + def test_meta_in_provider_list(self): + from litellm import LlmProviders + + assert hasattr(LlmProviders, "META") + assert LlmProviders.META.value == "meta" + assert "meta" in litellm.provider_list + + def test_meta_json_config_exists(self): + from litellm.llms.openai_like.json_loader import JSONProviderRegistry + + assert JSONProviderRegistry.exists("meta") + + meta = JSONProviderRegistry.get("meta") + assert meta is not None + assert meta.base_url == "https://api.meta.ai/v1" + assert meta.api_key_env == "META_API_KEY" + assert meta.api_base_env == "META_API_BASE" + + def test_meta_supports_responses_api(self): + from litellm.llms.openai_like.json_loader import JSONProviderRegistry + + assert JSONProviderRegistry.supports_responses_api("meta") + + def test_meta_in_openai_compatible_providers(self): + from litellm.constants import openai_compatible_providers + + assert "meta" in openai_compatible_providers + + def test_meta_provider_resolution(self): + from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider + + model, provider, api_key, api_base = get_llm_provider( + model="meta/muse-spark-1.1", + custom_llm_provider=None, + api_base=None, + api_key="sk-test", + ) + + assert model == "muse-spark-1.1" + assert provider == "meta" + assert api_base == "https://api.meta.ai/v1" + + def test_meta_api_base_override(self): + from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider + + model, provider, api_key, api_base = get_llm_provider( + model="meta/muse-spark-1.1", + custom_llm_provider=None, + api_base="https://custom.meta.ai/v1", + api_key="sk-test", + ) + + assert provider == "meta" + assert api_base == "https://custom.meta.ai/v1" + assert api_key == "sk-test" + + def test_meta_url_autodetection(self): + from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider + + model, provider, api_key, api_base = get_llm_provider( + model="muse-spark-1.1", + custom_llm_provider=None, + api_base="https://api.meta.ai/v1", + api_key=None, + ) + assert provider == "meta" + assert api_base == "https://api.meta.ai/v1" + + def test_meta_router_config(self): + from litellm import Router + + router = Router( + model_list=[ + { + "model_name": "muse-spark", + "litellm_params": { + "model": "meta/muse-spark-1.1", + "api_key": "test-key", + }, + } + ] + ) + + assert len(router.model_list) == 1 + assert router.model_list[0]["model_name"] == "muse-spark" + + +class TestMetaReasoningParams: + def test_muse_spark_supports_reasoning_effort(self): + params = litellm.get_supported_openai_params( + model="muse-spark-1.1", custom_llm_provider="meta" + ) + assert params is not None + assert "reasoning_effort" in params + + def test_reasoning_effort_mapped_through(self): + cfg = litellm.ProviderConfigManager.get_provider_chat_config( + model="muse-spark-1.1", provider=litellm.LlmProviders.META + ) + assert cfg is not None + mapped = cfg.map_openai_params( + non_default_params={"reasoning_effort": "xhigh"}, + optional_params={}, + model="muse-spark-1.1", + drop_params=False, + ) + assert mapped["reasoning_effort"] == "xhigh" + + def test_reasoning_effort_gated_on_capability(self): + """A meta model without reasoning metadata must not advertise reasoning_effort.""" + params = litellm.get_supported_openai_params( + model="some-non-reasoning-model", custom_llm_provider="meta" + ) + assert params is not None + assert "reasoning_effort" not in params + + +class TestMetaAnthropicMessages: + def test_meta_resolves_native_messages_config(self): + from litellm.llms.openai_like.messages.transformation import ( + JSONProviderAnthropicMessagesConfig, + ) + + cfg = litellm.ProviderConfigManager.get_provider_anthropic_messages_config( + model="muse-spark-1.1", provider=litellm.LlmProviders.META + ) + assert isinstance(cfg, JSONProviderAnthropicMessagesConfig) + + def test_json_provider_without_messages_endpoint_resolves_none(self): + cfg = litellm.ProviderConfigManager.get_provider_anthropic_messages_config( + model="some-model", provider=litellm.LlmProviders.PINSTRIPES + ) + assert cfg is None + + def test_complete_url_defaults_to_meta_base(self): + from litellm.llms.openai_like.json_loader import JSONProviderRegistry + from litellm.llms.openai_like.messages.transformation import ( + JSONProviderAnthropicMessagesConfig, + ) + + provider = JSONProviderRegistry.get("meta") + assert provider is not None + cfg = JSONProviderAnthropicMessagesConfig(provider) + + url = cfg.get_complete_url( + api_base=None, + api_key="sk-test", + model="muse-spark-1.1", + optional_params={}, + litellm_params={}, + ) + assert url == "https://api.meta.ai/v1/messages" + + override_url = cfg.get_complete_url( + api_base="https://custom.meta.ai/v1", + api_key="sk-test", + model="muse-spark-1.1", + optional_params={}, + litellm_params={}, + ) + assert override_url == "https://custom.meta.ai/v1/messages" + + def test_api_key_resolved_from_env(self, monkeypatch): + from litellm.llms.openai_like.json_loader import JSONProviderRegistry + from litellm.llms.openai_like.messages.transformation import ( + JSONProviderAnthropicMessagesConfig, + ) + + monkeypatch.setenv("META_API_KEY", "sk-env-key") + provider = JSONProviderRegistry.get("meta") + assert provider is not None + cfg = JSONProviderAnthropicMessagesConfig(provider) + + headers, _ = cfg.validate_anthropic_messages_environment( + headers={}, + model="muse-spark-1.1", + messages=[{"role": "user", "content": "hi"}], + optional_params={}, + litellm_params={}, + api_key=None, + api_base=None, + ) + assert headers["authorization"] == "Bearer sk-env-key" + assert headers["anthropic-version"] == "2023-06-01" + + +class TestMuseSparkModelInfo: + def test_muse_spark_pricing_and_capabilities(self): + info = litellm.get_model_info("meta/muse-spark-1.1") + + assert info["litellm_provider"] == "meta" + assert info["input_cost_per_token"] == 1.25e-06 + assert info["output_cost_per_token"] == 4.25e-06 + assert info["cache_read_input_token_cost"] == 1.5e-07 + assert info["max_input_tokens"] == 1048576 + assert info["supports_reasoning"] is True + assert info["supports_web_search"] is True + assert info["supports_vision"] is True + assert info["supports_function_calling"] is True + assert info["supports_prompt_caching"] is True + + def test_muse_spark_cost_calculation(self): + from litellm import completion_cost + from litellm.types.utils import ModelResponse, Usage + + response = ModelResponse( + model="muse-spark-1.1", + usage=Usage(prompt_tokens=1000, completion_tokens=500, total_tokens=1500), + ) + cost = completion_cost( + completion_response=response, + model="meta/muse-spark-1.1", + custom_llm_provider="meta", + ) + expected = 1000 * 1.25e-06 + 500 * 4.25e-06 + assert abs(cost - expected) < 1e-12 diff --git a/tests/test_litellm/ocr/test_ocr_azure_document_intelligence_api_base.py b/tests/test_litellm/ocr/test_ocr_azure_document_intelligence_api_base.py new file mode 100644 index 00000000000..0c8b1cc2836 --- /dev/null +++ b/tests/test_litellm/ocr/test_ocr_azure_document_intelligence_api_base.py @@ -0,0 +1,85 @@ +""" +Regression tests for Azure Document Intelligence api_base resolution in OCR. + +`azure_ai` exposes two OCR services on one provider; the `doc-intelligence` +sub-route must resolve to `AZURE_DOCUMENT_INTELLIGENCE_ENDPOINT`, not to the +generic `AZURE_AI_API_BASE` fallback that `get_llm_provider` injects. These tests +pin that routing and guard the backwards-compatibility contract that an explicitly +supplied api_base is always honoured. +""" + +from litellm.llms.azure_ai.ocr.common_utils import ( + is_azure_document_intelligence_model, +) +from litellm.ocr.main import _prepare_ocr_request, _rust_bridge_api_base + +_DOC = {"type": "document_url", "document_url": "https://example.com/doc.pdf"} +_DOC_INTELLIGENCE_ENDPOINT = "https://di.cognitiveservices.azure.com" +_AZURE_AI_API_BASE = "https://generic-azure-ai.example.com" + + +class _FakeLogging: + def update_from_kwargs(self, **kwargs: object) -> None: + return None + + +def _resolve_secret(name: str) -> str | None: + return { + "AZURE_DOCUMENT_INTELLIGENCE_ENDPOINT": _DOC_INTELLIGENCE_ENDPOINT, + "AZURE_AI_API_BASE": _AZURE_AI_API_BASE, + }.get(name) + + +def _prepare(model: str, api_base: str | None): + return _prepare_ocr_request( + model=model, + document=dict(_DOC), + api_key="test-key", + api_base=api_base, + timeout=None, + custom_llm_provider=None, + extra_headers=None, + kwargs={"litellm_logging_obj": _FakeLogging()}, + ) + + +class TestIsAzureDocumentIntelligenceModel: + def test_matches_doc_intelligence_route(self): + assert is_azure_document_intelligence_model("doc-intelligence/prebuilt-layout") + + def test_matches_documentintelligence_and_is_case_insensitive(self): + assert is_azure_document_intelligence_model("azure_ai/DocumentIntelligence/x") + + def test_does_not_match_mistral_route(self): + assert not is_azure_document_intelligence_model("mistral-document-ai-2505") + + +class TestDocIntelligenceApiBaseResolution: + def test_generic_azure_ai_base_does_not_hijack_doc_intelligence(self, monkeypatch): + """Without an explicit api_base, the AZURE_AI_API_BASE fallback must not + overwrite the endpoint, so it resolves to the Document Intelligence one.""" + monkeypatch.setenv("AZURE_AI_API_BASE", _AZURE_AI_API_BASE) + monkeypatch.delenv("AZURE_DOCUMENT_INTELLIGENCE_ENDPOINT", raising=False) + + prepared = _prepare("azure_ai/doc-intelligence/prebuilt-layout", None) + + assert prepared.api_base is None + assert _rust_bridge_api_base(prepared, _resolve_secret) == _DOC_INTELLIGENCE_ENDPOINT + + def test_explicit_api_base_is_honoured_for_doc_intelligence(self, monkeypatch): + """A caller-supplied api_base must always win, even for doc-intelligence.""" + monkeypatch.setenv("AZURE_AI_API_BASE", _AZURE_AI_API_BASE) + + custom = "https://my-di.cognitiveservices.azure.com" + prepared = _prepare("azure_ai/doc-intelligence/prebuilt-layout", custom) + + assert prepared.api_base == custom + assert _rust_bridge_api_base(prepared, _resolve_secret) == custom + + def test_generic_azure_ai_base_still_applies_to_mistral_ocr(self, monkeypatch): + """Non doc-intelligence azure_ai models keep using AZURE_AI_API_BASE.""" + monkeypatch.setenv("AZURE_AI_API_BASE", _AZURE_AI_API_BASE) + + prepared = _prepare("azure_ai/mistral-document-ai-2505", None) + + assert prepared.api_base == _AZURE_AI_API_BASE diff --git a/tests/test_litellm/proxy/_experimental/mcp_server/auth/test_user_api_key_auth_mcp.py b/tests/test_litellm/proxy/_experimental/mcp_server/auth/test_user_api_key_auth_mcp.py index 3db0f8540f9..ebaa6bc7cc0 100644 --- a/tests/test_litellm/proxy/_experimental/mcp_server/auth/test_user_api_key_auth_mcp.py +++ b/tests/test_litellm/proxy/_experimental/mcp_server/auth/test_user_api_key_auth_mcp.py @@ -1,14 +1,13 @@ import json import os import sys -from unittest.mock import AsyncMock, MagicMock, call as mock_call, patch +from unittest.mock import AsyncMock, MagicMock, patch import pytest +from fastapi import HTTPException from fastapi.testclient import TestClient -sys.path.insert( - 0, os.path.abspath("../../../..") -) # Adds the parent directory to the system path +sys.path.insert(0, os.path.abspath("../../../..")) # Adds the parent directory to the system path from starlette.datastructures import Headers @@ -78,20 +77,14 @@ class TestMCPRequestHandler: ) # Mock the helper methods instead of database calls - with patch.object( - MCPRequestHandler, "_get_allowed_mcp_servers_for_key" - ) as mock_key_servers: - with patch.object( - MCPRequestHandler, "_get_allowed_mcp_servers_for_team" - ) as mock_team_servers: + with patch.object(MCPRequestHandler, "_get_allowed_mcp_servers_for_key") as mock_key_servers: + with patch.object(MCPRequestHandler, "_get_allowed_mcp_servers_for_team") as mock_team_servers: # Set up return values mock_key_servers.return_value = key_servers mock_team_servers.return_value = team_servers # Call the method - result = await MCPRequestHandler.get_allowed_mcp_servers( - user_api_key_auth=mock_user_auth - ) + result = await MCPRequestHandler.get_allowed_mcp_servers(user_api_key_auth=mock_user_auth) # Assert the result (order-independent comparison) assert sorted(result) == sorted(expected_result) @@ -148,20 +141,14 @@ class TestMCPRequestHandler: ) # Mock the helper functions - with patch.object( - MCPRequestHandler, "_get_allowed_mcp_servers_for_key" - ) as mock_key_servers: - with patch.object( - MCPRequestHandler, "_get_allowed_mcp_servers_for_team" - ) as mock_team_servers: + with patch.object(MCPRequestHandler, "_get_allowed_mcp_servers_for_key") as mock_key_servers: + with patch.object(MCPRequestHandler, "_get_allowed_mcp_servers_for_team") as mock_team_servers: # Configure mocks to return the test data mock_key_servers.return_value = key_servers mock_team_servers.return_value = team_servers # Call the method - result = await MCPRequestHandler.get_allowed_mcp_servers( - user_api_key_auth - ) + result = await MCPRequestHandler.get_allowed_mcp_servers(user_api_key_auth) # Assert the result (order-independent comparison) assert sorted(result) == sorted(expected_servers) @@ -186,9 +173,7 @@ class TestMCPRequestHandler: ): """The require_key_mcp_access_defined general setting flips an empty key from inheriting its team's MCP servers (default) to inheriting none.""" - auth = UserAPIKeyAuth( - api_key="test-key", user_id="test-user", team_id="test-team" - ) + auth = UserAPIKeyAuth(api_key="test-key", user_id="test-user", team_id="test-team") with ( patch.object( MCPRequestHandler, @@ -272,23 +257,22 @@ class TestMCPRequestHandler: async def test_no_mcp_servers_sentinel_returns_empty(self, team_servers): """A key scoped to the no-mcp-servers sentinel resolves to zero servers, overriding team inheritance and never leaking the sentinel marker.""" - user_api_key_auth = UserAPIKeyAuth( - api_key="test-key", user_id="test-user", team_id="test-team" - ) + user_api_key_auth = UserAPIKeyAuth(api_key="test-key", user_id="test-user", team_id="test-team") key_object_permission = MagicMock() - key_object_permission.mcp_servers = [ - SpecialMCPServerNames.no_mcp_servers.value - ] + key_object_permission.mcp_servers = [SpecialMCPServerNames.no_mcp_servers.value] - with patch.object( - MCPRequestHandler, - "_get_key_object_permission", - return_value=key_object_permission, - ), patch.object( - MCPRequestHandler, - "_get_allowed_mcp_servers_for_team", - new_callable=AsyncMock, - return_value=team_servers, + with ( + patch.object( + MCPRequestHandler, + "_get_key_object_permission", + return_value=key_object_permission, + ), + patch.object( + MCPRequestHandler, + "_get_allowed_mcp_servers_for_team", + new_callable=AsyncMock, + return_value=team_servers, + ), ): result = await MCPRequestHandler.get_allowed_mcp_servers(user_api_key_auth) @@ -309,9 +293,7 @@ class TestMCPRequestHandler: "_get_key_object_permission", return_value=key_object_permission, ): - result = await MCPRequestHandler._get_allowed_mcp_servers_for_key( - user_api_key_auth - ) + result = await MCPRequestHandler._get_allowed_mcp_servers_for_key(user_api_key_auth) assert result == [SpecialMCPServerNames.no_mcp_servers.value] @@ -320,9 +302,7 @@ class TestMCPRequestHandler: # Test case: None values in database mock_prisma_client = MagicMock() - mock_prisma_client.db.litellm_objectpermissiontable.find_unique.return_value = ( - None - ) + mock_prisma_client.db.litellm_objectpermissiontable.find_unique.return_value = None mock_prisma_client.db.litellm_teamtable.find_unique.return_value = None user_api_key_auth = UserAPIKeyAuth( @@ -337,9 +317,7 @@ class TestMCPRequestHandler: assert result == [] # Test case: Exception handling - mock_prisma_client.db.litellm_objectpermissiontable.find_unique.side_effect = ( - Exception("DB Error") - ) + mock_prisma_client.db.litellm_objectpermissiontable.find_unique.side_effect = Exception("DB Error") with patch("litellm.proxy.proxy_server.prisma_client", mock_prisma_client): result = await MCPRequestHandler.get_allowed_mcp_servers(user_api_key_auth) @@ -384,15 +362,9 @@ class TestMCPRequestHandler: access_group_ids=["grp-mcp"], ) with ( - patch.object( - MCPRequestHandler, "_get_allowed_mcp_servers_for_key" - ) as mock_key, - patch.object( - MCPRequestHandler, "_get_allowed_mcp_servers_for_team" - ) as mock_team, - patch.object( - MCPRequestHandler, "_get_key_access_group_mcp_server_extras" - ) as mock_grants, + patch.object(MCPRequestHandler, "_get_allowed_mcp_servers_for_key") as mock_key, + patch.object(MCPRequestHandler, "_get_allowed_mcp_servers_for_team") as mock_team, + patch.object(MCPRequestHandler, "_get_key_access_group_mcp_server_extras") as mock_grants, ): mock_key.return_value = key_servers mock_team.return_value = team_servers @@ -414,13 +386,9 @@ class TestMCPRequestHandler: "litellm.proxy.auth.auth_checks._get_mcp_server_ids_from_access_groups", new=AsyncMock(return_value=[]), ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): - result = await MCPRequestHandler._get_key_access_group_mcp_server_extras( - auth - ) + result = await MCPRequestHandler._get_key_access_group_mcp_server_extras(auth) assert result == [] # expand_permission_list must not be reached when there are no raw ids. mock_mgr.expand_permission_list.assert_not_called() @@ -433,14 +401,10 @@ class TestMCPRequestHandler: "litellm.proxy.auth.auth_checks._get_mcp_server_ids_from_access_groups", new=AsyncMock(return_value=["alias-a", "srv-b"]), ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.expand_permission_list.return_value = ["srv-a", "srv-b"] - result = await MCPRequestHandler._get_key_access_group_mcp_server_extras( - auth - ) + result = await MCPRequestHandler._get_key_access_group_mcp_server_extras(auth) assert sorted(result) == ["srv-a", "srv-b"] mock_mgr.expand_permission_list.assert_called_once_with(["alias-a", "srv-b"]) @@ -451,9 +415,7 @@ class TestMCPRequestHandler: "litellm.proxy.auth.auth_checks._get_mcp_server_ids_from_access_groups", new=AsyncMock(side_effect=Exception("db down")), ): - result = await MCPRequestHandler._get_key_access_group_mcp_server_extras( - auth - ) + result = await MCPRequestHandler._get_key_access_group_mcp_server_extras(auth) assert result == [] @pytest.mark.parametrize( @@ -743,9 +705,7 @@ class TestMCPRequestHandler: # Verify MCP servers mcp_servers_header = extracted_headers.get(SpecialHeaders.mcp_servers.value) mcp_servers = None - if ( - mcp_servers_header is not None - ): # Changed from 'if mcp_servers_header:' to handle empty strings + if mcp_servers_header is not None: # Changed from 'if mcp_servers_header:' to handle empty strings try: # First try to parse as JSON array for backward compatibility try: @@ -754,16 +714,12 @@ class TestMCPRequestHandler: mcp_servers = None except (json.JSONDecodeError, TypeError, ValueError): # If JSON parsing fails, treat as comma-separated list - mcp_servers = [ - s.strip() for s in mcp_servers_header.split(",") if s.strip() - ] + mcp_servers = [s.strip() for s in mcp_servers_header.split(",") if s.strip()] except Exception: mcp_servers = None # If we got an empty string or parsing resulted in no servers, return empty list - if mcp_servers_header == "" or ( - mcp_servers is not None and len(mcp_servers) == 0 - ): + if mcp_servers_header == "" or (mcp_servers is not None and len(mcp_servers) == 0): mcp_servers = [] assert mcp_servers == expected_result["mcp_servers"] @@ -795,6 +751,40 @@ class TestMCPRequestHandler: # For these tests, mcp_server_auth_headers should be empty assert mcp_server_auth_headers == {} + def test_duplicate_authorization_header_is_rejected(self): + """A request carrying more than one Authorization header is malformed for bearer auth and, + for the client-forwarded token modes, would make which upstream token is forwarded ambiguous. + The ingress header converter must reject it with a 400 rather than silently keeping one.""" + scope = { + "type": "http", + "method": "POST", + "path": "/mcp/tp_server", + "headers": [ + (b"authorization", b"Bearer upstream-token-a"), + (b"authorization", b"Bearer upstream-token-b"), + (b"content-type", b"application/json"), + ], + } + with pytest.raises(HTTPException) as exc_info: + MCPRequestHandler._safe_get_headers_from_scope(scope) + assert exc_info.value.status_code == 400 + assert "Authorization" in str(exc_info.value.detail) + + def test_single_authorization_header_is_forwarded_verbatim(self): + """The rejection must not disturb the normal single-Authorization case: the value passes + through unchanged (guards against the duplicate check over-matching).""" + scope = { + "type": "http", + "method": "POST", + "path": "/mcp/tp_server", + "headers": [ + (b"authorization", b"Bearer upstream-token"), + (b"content-type", b"application/json"), + ], + } + headers = MCPRequestHandler._safe_get_headers_from_scope(scope) + assert headers.get("authorization") == "Bearer upstream-token" + @pytest.mark.asyncio class TestMCPOAuth2AuthFlow: @@ -833,9 +823,7 @@ class TestMCPOAuth2AuthFlow: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", new_callable=AsyncMock, ) as mock_auth, - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = oauth2_server ( @@ -851,10 +839,7 @@ class TestMCPOAuth2AuthFlow: # The upstream token is never validated as a LiteLLM key ... mock_auth.assert_not_called() # ... and is preserved for upstream forwarding. - assert ( - oauth2_headers.get("Authorization") - == "Bearer atlassian-oauth2-access-token-xyz" - ) + assert oauth2_headers.get("Authorization") == "Bearer atlassian-oauth2-access-token-xyz" async def test_explicit_litellm_key_with_oauth2_authorization(self): """ @@ -893,9 +878,7 @@ class TestMCPOAuth2AuthFlow: assert call_args.kwargs["api_key"] == "sk-litellm-valid-key" # OAuth2 headers should still contain the Authorization token - assert ( - oauth2_headers.get("Authorization") == "Bearer atlassian-oauth2-token" - ) + assert oauth2_headers.get("Authorization") == "Bearer atlassian-oauth2-token" async def test_litellm_key_in_authorization_backward_compat(self): """ @@ -997,9 +980,7 @@ class TestMCPOAuth2AuthFlow: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_proxy_exception, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = oauth2_server with pytest.raises(ProxyException) as exc_info: @@ -1068,9 +1049,7 @@ class TestMCPPublicRouteGuard: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): # Explicit unresolvable target — proves auth still fails even # when the registry has no info to fall back to. @@ -1101,9 +1080,7 @@ class TestMCPPublicRouteGuard: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = None with pytest.raises(HTTPException) as exc_info: @@ -1157,9 +1134,7 @@ class TestMCPPassthroughColdStartAdmission: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, patch( "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp._is_mcp_passthrough_cold_start" ) as mock_cold_start, @@ -1199,9 +1174,7 @@ class TestMCPPassthroughColdStartAdmission: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = ( TestMCPPassthroughColdStartAdmission._make_passthrough_server() @@ -1229,9 +1202,7 @@ class TestMCPPassthroughColdStartAdmission: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = ( TestMCPPassthroughColdStartAdmission._make_passthrough_server() @@ -1263,18 +1234,14 @@ class TestMCPPassthroughColdStartAdmission: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.IPAddressUtils.get_mcp_client_ip", return_value="203.0.113.10", ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = None with pytest.raises(HTTPException) as exc_info: await MCPRequestHandler.process_mcp_request(scope) assert exc_info.value.status_code == 401 - mock_mgr.get_mcp_server_by_name.assert_any_call( - "passthrough_server", client_ip="203.0.113.10" - ) + mock_mgr.get_mcp_server_by_name.assert_any_call("passthrough_server", client_ip="203.0.113.10") async def test_cold_start_propagates_non_401_http_error(self): from fastapi import HTTPException @@ -1294,9 +1261,7 @@ class TestMCPPassthroughColdStartAdmission: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_forbidden, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = ( TestMCPPassthroughColdStartAdmission._make_passthrough_server() @@ -1329,9 +1294,7 @@ class TestMCPPassthroughColdStartAdmission: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_server_error, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = ( TestMCPPassthroughColdStartAdmission._make_passthrough_server() @@ -1357,9 +1320,7 @@ class TestMCPPassthroughColdStartAdmission: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = ( TestMCPPassthroughColdStartAdmission._make_passthrough_server() @@ -1367,9 +1328,7 @@ class TestMCPPassthroughColdStartAdmission: auth_result, *_rest = await MCPRequestHandler.process_mcp_request(scope) assert isinstance(auth_result, UserAPIKeyAuth) - mock_mgr.get_mcp_server_by_name.assert_any_call( - "passthrough_server", client_ip="" - ) + mock_mgr.get_mcp_server_by_name.assert_any_call("passthrough_server", client_ip="") async def test_cold_start_allows_proxy_exception_401_for_path_target(self): from litellm.proxy._types import ProxyException @@ -1394,9 +1353,7 @@ class TestMCPPassthroughColdStartAdmission: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = ( TestMCPPassthroughColdStartAdmission._make_passthrough_server() @@ -1404,9 +1361,7 @@ class TestMCPPassthroughColdStartAdmission: auth_result, *_rest = await MCPRequestHandler.process_mcp_request(scope) assert isinstance(auth_result, UserAPIKeyAuth) - mock_mgr.get_mcp_server_by_name.assert_any_call( - "passthrough_server", client_ip="" - ) + mock_mgr.get_mcp_server_by_name.assert_any_call("passthrough_server", client_ip="") @pytest.mark.asyncio @@ -1450,12 +1405,10 @@ class TestMCPOAuth2FallbackTargetGating: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): - mock_mgr.get_mcp_server_by_name.return_value = ( - TestMCPOAuth2FallbackTargetGating._make_server(MCPAuth.api_key) + mock_mgr.get_mcp_server_by_name.return_value = TestMCPOAuth2FallbackTargetGating._make_server( + MCPAuth.api_key ) with pytest.raises(HTTPException) as exc_info: await MCPRequestHandler.process_mcp_request(scope) @@ -1483,9 +1436,7 @@ class TestMCPOAuth2FallbackTargetGating: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = None with pytest.raises(HTTPException) as exc_info: @@ -1523,12 +1474,10 @@ class TestMCPOAuth2FallbackTargetGating: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): - mock_mgr.get_mcp_server_by_name.return_value = ( - TestMCPOAuth2FallbackTargetGating._make_server(MCPAuth.oauth2) + mock_mgr.get_mcp_server_by_name.return_value = TestMCPOAuth2FallbackTargetGating._make_server( + MCPAuth.oauth2 ) with pytest.raises(HTTPException) as exc_info: await MCPRequestHandler.process_mcp_request(scope) @@ -1562,15 +1511,11 @@ class TestMCPOAuth2FallbackTargetGating: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): - mock_mgr.get_mcp_server_by_name.return_value = ( - TestMCPOAuth2FallbackTargetGating._make_server( - auth_type=MCPAuth.none, - is_oauth_passthrough=True, - ) + mock_mgr.get_mcp_server_by_name.return_value = TestMCPOAuth2FallbackTargetGating._make_server( + auth_type=MCPAuth.none, + is_oauth_passthrough=True, ) auth_result, *_rest = await MCPRequestHandler.process_mcp_request(scope) assert isinstance(auth_result, UserAPIKeyAuth) @@ -1598,9 +1543,7 @@ class TestMCPOAuth2FallbackTargetGating: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.IPAddressUtils.get_mcp_client_ip", return_value="203.0.113.10", ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = None with pytest.raises(HTTPException) as exc_info: @@ -1612,9 +1555,7 @@ class TestMCPOAuth2FallbackTargetGating: # resolve to ``None`` (hidden by client IP) so neither bypass # opens. Use ``assert_any_call`` to assert the IP-scoped lookup # happened without locking the count. - mock_mgr.get_mcp_server_by_name.assert_any_call( - "hidden_oauth2_server", client_ip="203.0.113.10" - ) + mock_mgr.get_mcp_server_by_name.assert_any_call("hidden_oauth2_server", client_ip="203.0.113.10") async def test_fallback_blocked_when_any_target_in_header_is_not_oauth2(self): """ @@ -1649,9 +1590,7 @@ class TestMCPOAuth2FallbackTargetGating: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.side_effect = mock_lookup with pytest.raises(HTTPException) as exc_info: @@ -1732,17 +1671,10 @@ class TestMCPDelegateAuthToUpstream: delegate_auth_to_upstream=True, available_on_public_internet=True, ) - assert ( - manager._build_mcp_server_table(delegated).delegate_auth_to_upstream is True - ) + assert manager._build_mcp_server_table(delegated).delegate_auth_to_upstream is True - not_delegated = delegated.model_copy( - update={"delegate_auth_to_upstream": False} - ) - assert ( - manager._build_mcp_server_table(not_delegated).delegate_auth_to_upstream - is False - ) + not_delegated = delegated.model_copy(update={"delegate_auth_to_upstream": False}) + assert manager._build_mcp_server_table(not_delegated).delegate_auth_to_upstream is False def test_build_mcp_server_table_preserves_oauth_passthrough(self): """Registry → API list rows must expose oauth_passthrough for the UI. @@ -1773,9 +1705,7 @@ class TestMCPDelegateAuthToUpstream: assert row.delegate_auth_to_upstream is False not_passthrough = passthrough.model_copy(update={"oauth_passthrough": False}) - assert ( - manager._build_mcp_server_table(not_passthrough).oauth_passthrough is False - ) + assert manager._build_mcp_server_table(not_passthrough).oauth_passthrough is False async def test_delegate_skips_litellm_auth_with_no_authorization(self): """ @@ -1796,15 +1726,11 @@ class TestMCPDelegateAuthToUpstream: patch( "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", ) as mock_auth, - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): - mock_mgr.get_mcp_server_by_name.return_value = ( - TestMCPDelegateAuthToUpstream._make_server( - auth_type=MCPAuth.oauth2, - delegate_auth_to_upstream=True, - ) + mock_mgr.get_mcp_server_by_name.return_value = TestMCPDelegateAuthToUpstream._make_server( + auth_type=MCPAuth.oauth2, + delegate_auth_to_upstream=True, ) auth_result, *_rest = await MCPRequestHandler.process_mcp_request(scope) assert isinstance(auth_result, UserAPIKeyAuth) @@ -1834,15 +1760,11 @@ class TestMCPDelegateAuthToUpstream: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", new_callable=AsyncMock, ) as mock_auth, - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): - mock_mgr.get_mcp_server_by_name.return_value = ( - TestMCPDelegateAuthToUpstream._make_server( - auth_type=MCPAuth.oauth2, - delegate_auth_to_upstream=True, - ) + mock_mgr.get_mcp_server_by_name.return_value = TestMCPDelegateAuthToUpstream._make_server( + auth_type=MCPAuth.oauth2, + delegate_auth_to_upstream=True, ) ( auth_result, @@ -1880,15 +1802,11 @@ class TestMCPDelegateAuthToUpstream: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): - mock_mgr.get_mcp_server_by_name.return_value = ( - TestMCPDelegateAuthToUpstream._make_server( - auth_type=MCPAuth.oauth2, - delegate_auth_to_upstream=False, - ) + mock_mgr.get_mcp_server_by_name.return_value = TestMCPDelegateAuthToUpstream._make_server( + auth_type=MCPAuth.oauth2, + delegate_auth_to_upstream=False, ) with pytest.raises(HTTPException) as exc_info: await MCPRequestHandler.process_mcp_request(scope) @@ -1919,15 +1837,11 @@ class TestMCPDelegateAuthToUpstream: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): - mock_mgr.get_mcp_server_by_name.return_value = ( - TestMCPDelegateAuthToUpstream._make_server( - auth_type=MCPAuth.api_key, - delegate_auth_to_upstream=True, - ) + mock_mgr.get_mcp_server_by_name.return_value = TestMCPDelegateAuthToUpstream._make_server( + auth_type=MCPAuth.api_key, + delegate_auth_to_upstream=True, ) with pytest.raises(HTTPException) as exc_info: await MCPRequestHandler.process_mcp_request(scope) @@ -1971,9 +1885,7 @@ class TestMCPDelegateAuthToUpstream: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.side_effect = mock_lookup with pytest.raises(HTTPException) as exc_info: @@ -2003,9 +1915,7 @@ class TestMCPDelegateAuthToUpstream: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_user_api_key_auth_fails, ), - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = None with pytest.raises(HTTPException) as exc_info: @@ -2034,15 +1944,11 @@ class TestMCPDelegateAuthToUpstream: new_callable=AsyncMock, return_value=UserAPIKeyAuth(user_id="real-user"), ) as mock_auth, - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): - mock_mgr.get_mcp_server_by_name.return_value = ( - TestMCPDelegateAuthToUpstream._make_server( - auth_type=MCPAuth.oauth2, - delegate_auth_to_upstream=True, - ) + mock_mgr.get_mcp_server_by_name.return_value = TestMCPDelegateAuthToUpstream._make_server( + auth_type=MCPAuth.oauth2, + delegate_auth_to_upstream=True, ) auth_result, *_rest = await MCPRequestHandler.process_mcp_request(scope) assert isinstance(auth_result, UserAPIKeyAuth) @@ -2074,15 +1980,11 @@ class TestMCPDelegateAuthToUpstream: new_callable=AsyncMock, return_value=UserAPIKeyAuth(user_id="real-user"), ) as mock_auth, - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): - mock_mgr.get_mcp_server_by_name.return_value = ( - TestMCPDelegateAuthToUpstream._make_server( - auth_type=MCPAuth.oauth2, - delegate_auth_to_upstream=True, - ) + mock_mgr.get_mcp_server_by_name.return_value = TestMCPDelegateAuthToUpstream._make_server( + auth_type=MCPAuth.oauth2, + delegate_auth_to_upstream=True, ) ( auth_result, @@ -2135,9 +2037,7 @@ class TestMCPDelegateAuthToUpstream: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_auth_raises, ) as mock_auth, - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = m2m_server # No delegate bypass → normal auth is attempted → 401 raised @@ -2188,9 +2088,7 @@ class TestMCPDelegateAuthToUpstream: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_auth_raises, ) as mock_auth, - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = legacy_m2m_server with pytest.raises(HTTPException) as exc_info: @@ -2235,9 +2133,7 @@ class TestMCPDelegateAuthToUpstream: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_auth_raises, ) as mock_auth, - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = pkce_server auth, *_rest = await MCPRequestHandler.process_mcp_request(scope) @@ -2278,9 +2174,7 @@ class TestMCPDelegateAuthToUpstream: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_auth_raises, ) as mock_auth, - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.return_value = internal_server auth, *_rest = await MCPRequestHandler.process_mcp_request(scope) @@ -2426,6 +2320,104 @@ class TestMCPDelegateAuthToUpstream: assert "public-server" in result assert "internal-server" in result + async def test_true_passthrough_skips_litellm_auth_anonymously(self): + """auth_type=true_passthrough performs no admission auth: the caller's Authorization is an + upstream token forwarded unchanged and user_api_key_auth is never called.""" + from litellm.types.mcp import MCPAuth + + scope = { + "type": "http", + "method": "POST", + "path": "/mcp/true_passthrough_server", + "headers": [(b"authorization", b"Bearer upstream-token")], + } + + with ( + patch( + "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", + new_callable=AsyncMock, + ) as mock_auth, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, + ): + mock_mgr.get_mcp_server_by_name.return_value = TestMCPDelegateAuthToUpstream._make_server( + auth_type=MCPAuth.true_passthrough, + ) + ( + auth_result, + _, + _, + _, + oauth2_headers, + _, + ) = await MCPRequestHandler.process_mcp_request(scope) + assert isinstance(auth_result, UserAPIKeyAuth) + assert auth_result.api_key is None + assert oauth2_headers.get("Authorization") == "Bearer upstream-token" + mock_auth.assert_not_called() + + async def test_true_passthrough_mixed_targets_fail_closed(self): + """One true_passthrough target mixed with a non-passthrough target must NOT skip admission.""" + from fastapi import HTTPException + + from litellm.types.mcp import MCPAuth + + scope = { + "type": "http", + "method": "POST", + "path": "/mcp", + "headers": [(b"x-mcp-servers", b"tp_server,plain_server")], + } + + async def mock_user_api_key_auth_fails(api_key, request): + raise HTTPException(status_code=401, detail="Invalid API key") + + def mock_lookup(name, client_ip=None): + if name == "tp_server": + return TestMCPDelegateAuthToUpstream._make_server( + auth_type=MCPAuth.true_passthrough, + ) + return TestMCPDelegateAuthToUpstream._make_server(auth_type=MCPAuth.api_key) + + with ( + patch( + "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", + side_effect=mock_user_api_key_auth_fails, + ), + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, + ): + mock_mgr.get_mcp_server_by_name.side_effect = mock_lookup + with pytest.raises(HTTPException) as exc_info: + await MCPRequestHandler.process_mcp_request(scope) + assert exc_info.value.status_code == 401 + + async def test_get_allowed_servers_includes_true_passthrough(self): + """Anonymous callers can reach true_passthrough servers; admission is delegated upstream.""" + from litellm.proxy._experimental.mcp_server.mcp_server_manager import ( + MCPServerManager, + ) + from litellm.types.mcp import MCPAuth + from litellm.types.mcp_server.mcp_server_manager import MCPServer + + manager = MCPServerManager() + tp_server = MCPServer( + server_id="tp-server", + name="tp_server", + transport="http", + auth_type=MCPAuth.true_passthrough, + available_on_public_internet=True, + ) + manager.registry = {tp_server.server_id: tp_server} + + with patch.object( + MCPRequestHandler, + "get_allowed_mcp_servers", + new_callable=AsyncMock, + return_value=[], + ): + result = await manager.get_allowed_mcp_servers(None) + + assert "tp-server" in result + def test_extract_target_server_names_matches_routing_parser(self): """ Regression: _extract_target_server_names_from_path must match the @@ -2464,13 +2456,12 @@ class TestMCPDelegateAuthToUpstream: ("/", []), ] for path_input, expected in cases: - assert ( - MCPRequestHandler._extract_target_server_names_from_path(path_input) - == expected - ), f"path={path_input!r} → expected {expected!r}" - assert ( - _get_mcp_servers_in_path(path_input) or [] - ) == expected, f"path={path_input!r} → routing expected {expected!r}" + assert MCPRequestHandler._extract_target_server_names_from_path(path_input) == expected, ( + f"path={path_input!r} → expected {expected!r}" + ) + assert (_get_mcp_servers_in_path(path_input) or []) == expected, ( + f"path={path_input!r} → routing expected {expected!r}" + ) async def test_delegate_does_not_bypass_on_extra_path_segment(self): """ @@ -2514,9 +2505,7 @@ class TestMCPDelegateAuthToUpstream: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_auth_raises, ) as mock_auth, - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.side_effect = lookup_by_name with pytest.raises(HTTPException) as exc_info: @@ -2579,9 +2568,7 @@ class TestMCPDelegateAuthToUpstream: "litellm.proxy._experimental.mcp_server.auth.user_api_key_auth_mcp.user_api_key_auth", side_effect=mock_auth_raises, ) as mock_auth, - patch( - "litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager" - ) as mock_mgr, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.global_mcp_server_manager") as mock_mgr, ): mock_mgr.get_mcp_server_by_name.side_effect = lookup_by_name # Bypass MUST NOT fire — path-derived target is the non-delegate @@ -2601,15 +2588,12 @@ class TestMCPDelegateAuthToUpstream: empty-list case, which fails closed). """ # Path matches /mcp/... — header is ignored. - assert MCPRequestHandler._resolve_target_server_names( - path="/mcp/foo", mcp_servers_header=["evil"] - ) == ["foo"] - assert MCPRequestHandler._resolve_target_server_names( - path="/mcp/foo,bar", mcp_servers_header=["evil"] - ) == ["foo", "bar"] - assert MCPRequestHandler._resolve_target_server_names( - path="/foo/mcp", mcp_servers_header=["evil"] - ) == ["foo"] + assert MCPRequestHandler._resolve_target_server_names(path="/mcp/foo", mcp_servers_header=["evil"]) == ["foo"] + assert MCPRequestHandler._resolve_target_server_names(path="/mcp/foo,bar", mcp_servers_header=["evil"]) == [ + "foo", + "bar", + ] + assert MCPRequestHandler._resolve_target_server_names(path="/foo/mcp", mcp_servers_header=["evil"]) == ["foo"] # Path does not match — header is trusted. assert MCPRequestHandler._resolve_target_server_names( path="/.well-known/oauth-authorization-server", @@ -2653,16 +2637,12 @@ class TestMCPCustomHeaderName: (None, "", "x-mcp-auth"), ], ) - def test_get_mcp_client_side_auth_header_name( - self, env_var, general_setting, expected_header_name - ): + def test_get_mcp_client_side_auth_header_name(self, env_var, general_setting, expected_header_name): """Test that custom header name configuration works correctly""" # Mock the secret manager and general settings with patch("litellm.secret_managers.main.get_secret_str") as mock_get_secret: - with patch( - "litellm.proxy.proxy_server.general_settings" - ) as mock_general_settings: + with patch("litellm.proxy.proxy_server.general_settings") as mock_general_settings: # Configure mocks mock_get_secret.return_value = env_var mock_general_settings.get.return_value = general_setting @@ -2685,9 +2665,7 @@ class TestMCPCustomHeaderName: if env_var is None: # When env var is None, general settings should be checked (twice if not None) expected_general_calls = 2 if general_setting is not None else 1 - assert ( - mock_general_settings.get.call_count == expected_general_calls - ) + assert mock_general_settings.get.call_count == expected_general_calls for call in mock_general_settings.get.call_args_list: assert call.args == ("mcp_client_side_auth_header_name",) else: @@ -2728,9 +2706,7 @@ class TestMCPCustomHeaderName: ), ], ) - def test_get_mcp_auth_header_from_headers_with_custom_name( - self, custom_header_name, headers, expected_auth_header - ): + def test_get_mcp_auth_header_from_headers_with_custom_name(self, custom_header_name, headers, expected_auth_header): """Test that MCP auth header extraction uses custom header name""" # Mock the header name method @@ -2749,9 +2725,7 @@ class TestMCPCustomHeaderName: extracted_headers = MCPRequestHandler._safe_get_headers_from_scope(scope) # Call the method - result = MCPRequestHandler._get_mcp_auth_header_from_headers( - extracted_headers - ) + result = MCPRequestHandler._get_mcp_auth_header_from_headers(extracted_headers) # Assert the result assert result == expected_auth_header @@ -2818,9 +2792,7 @@ class TestMCPCustomHeaderName: from starlette.datastructures import Headers # Test case 1: No server-specific headers - headers = Headers( - {"x-litellm-api-key": "test-key", "content-type": "application/json"} - ) + headers = Headers({"x-litellm-api-key": "test-key", "content-type": "application/json"}) result = MCPRequestHandler._get_mcp_server_auth_headers_from_headers(headers) assert result == {} @@ -2904,17 +2876,13 @@ class TestMCPCustomHeaderName: assert result == {"github_mcp": {"Authorization": "Bearer github-mcp-token"}} # Test case 8: Edge case - empty header value - headers = Headers( - {"x-litellm-api-key": "test-key", "x-mcp-github-authorization": ""} - ) + headers = Headers({"x-litellm-api-key": "test-key", "x-mcp-github-authorization": ""}) result = MCPRequestHandler._get_mcp_server_auth_headers_from_headers(headers) assert result == {"github": {"Authorization": ""}} # Test case 9: Edge case - very long header value long_token = "Bearer " + "x" * 1000 - headers = Headers( - {"x-litellm-api-key": "test-key", "x-mcp-github-authorization": long_token} - ) + headers = Headers({"x-litellm-api-key": "test-key", "x-mcp-github-authorization": long_token}) result = MCPRequestHandler._get_mcp_server_auth_headers_from_headers(headers) assert result == {"github": {"Authorization": long_token}} @@ -2980,9 +2948,7 @@ class TestMCPAccessGroupsE2E: # Assert the results assert auth_result.api_key == "test-api-key" assert mcp_auth_header is None - assert ( - mcp_servers is None - ) # x-mcp-access-groups is not parsed as mcp_servers + assert mcp_servers is None # x-mcp-access-groups is not parsed as mcp_servers assert mcp_server_auth_headers == {} # Verify the mock was called @@ -3097,9 +3063,7 @@ def test_mcp_path_based_server_segregation(monkeypatch): # Use TestClient to make a request to /mcp/zapier,group1/tools client = TestClient(app) - response = client.get( - "/mcp/zapier,group1/tools", headers={"x-litellm-api-key": "test"} - ) + response = client.get("/mcp/zapier,group1/tools", headers={"x-litellm-api-key": "test"}) assert response.status_code == 200 assert response.json() == {"status": "ok"} @@ -3177,15 +3141,11 @@ async def test_get_team_object_permission_with_already_loaded_permission(): mock_prisma, ): with patch("litellm.proxy.auth.auth_checks.get_team_object") as mock_get_team: - with patch( - "litellm.proxy.auth.auth_checks.get_object_permission" - ) as mock_get_perm: + with patch("litellm.proxy.auth.auth_checks.get_object_permission") as mock_get_perm: mock_get_team.return_value = mock_team_obj # Call the method - result = await MCPRequestHandler._get_team_object_permission( - mock_user_auth - ) + result = await MCPRequestHandler._get_team_object_permission(mock_user_auth) # Assert we got the object permission assert result == mock_object_permission @@ -3272,9 +3232,7 @@ async def test_get_team_object_permission_ui_session_team_skips_db_lookup(): mock_prisma = MagicMock() with patch("litellm.proxy.proxy_server.prisma_client", mock_prisma): with patch("litellm.proxy.auth.auth_checks.get_team_object") as mock_get_team: - result = await MCPRequestHandler._get_team_object_permission( - mock_user_auth - ) + result = await MCPRequestHandler._get_team_object_permission(mock_user_auth) assert result is None mock_get_team.assert_not_called() @@ -3342,9 +3300,7 @@ async def test_get_allowed_tools_for_server_ui_session_team_keeps_key_restrictio detail={"error": "Team doesn't exist in db. Team=litellm-dashboard."}, ), ): - with patch.object( - MCPRequestHandler, "_get_key_object_permission", return_value=key_perm - ): + with patch.object(MCPRequestHandler, "_get_key_object_permission", return_value=key_perm): result = await MCPRequestHandler.get_allowed_tools_for_server( server_id="server_1", user_api_key_auth=user_api_key_auth, @@ -3410,9 +3366,7 @@ async def test_get_allowed_mcp_servers_for_team_uses_helper(): return_value=["group-server1", "group-server2"], ) as mock_get_access_group_servers, ): - result = await MCPRequestHandler._get_allowed_mcp_servers_for_team( - mock_user_auth - ) + result = await MCPRequestHandler._get_allowed_mcp_servers_for_team(mock_user_auth) assert set(result) == { "direct-server1", @@ -3455,9 +3409,7 @@ async def test_get_allowed_mcp_servers_for_team_with_no_object_permission(): return_value=mock_team, ), ): - result = await MCPRequestHandler._get_allowed_mcp_servers_for_team( - mock_user_auth - ) + result = await MCPRequestHandler._get_allowed_mcp_servers_for_team(mock_user_auth) assert result == [] @@ -3507,9 +3459,7 @@ async def test_get_allowed_mcp_servers_for_team_without_team_id_returns_empty(): ), ], ) -async def test_get_allowed_mcp_servers_for_key_guard_conditions( - user_api_key_auth, prisma_client_value, scenario -): +async def test_get_allowed_mcp_servers_for_key_guard_conditions(user_api_key_auth, prisma_client_value, scenario): """Ensure guard clauses return [] before hitting get_object_permission.""" with patch( @@ -3517,9 +3467,7 @@ async def test_get_allowed_mcp_servers_for_key_guard_conditions( new_callable=AsyncMock, ) as mock_get_perm: with patch("litellm.proxy.proxy_server.prisma_client", prisma_client_value): - result = await MCPRequestHandler._get_allowed_mcp_servers_for_key( - user_api_key_auth - ) + result = await MCPRequestHandler._get_allowed_mcp_servers_for_key(user_api_key_auth) assert result == [] mock_get_perm.assert_not_called() @@ -3546,9 +3494,7 @@ async def test_get_allowed_mcp_servers_for_key_returns_empty_when_db_returns_non ): mock_get_perm.return_value = None - result = await MCPRequestHandler._get_allowed_mcp_servers_for_key( - user_api_key_auth - ) + result = await MCPRequestHandler._get_allowed_mcp_servers_for_key(user_api_key_auth) assert result == [] mock_get_perm.assert_awaited_once() @@ -3591,14 +3537,10 @@ async def test_get_allowed_mcp_servers_for_key_prefers_in_memory_permission(): "litellm.proxy.auth.auth_checks.get_object_permission", new_callable=AsyncMock, ) as mock_get_perm: - with patch.object( - MCPRequestHandler, "_get_mcp_servers_from_access_groups" - ) as mock_access_groups: + with patch.object(MCPRequestHandler, "_get_mcp_servers_from_access_groups") as mock_access_groups: mock_access_groups.return_value = ["group-server"] - result = await MCPRequestHandler._get_allowed_mcp_servers_for_key( - user_api_key_auth - ) + result = await MCPRequestHandler._get_allowed_mcp_servers_for_key(user_api_key_auth) assert set(result) == {"direct-server", "group-server"} mock_get_perm.assert_not_called() @@ -3619,21 +3561,13 @@ class TestAgentMCPPermissions: team_id="test-team", agent_id="agent-123", ) - with patch.object( - MCPRequestHandler, "_get_allowed_mcp_servers_for_key" - ) as mock_key: - with patch.object( - MCPRequestHandler, "_get_allowed_mcp_servers_for_team" - ) as mock_team: - with patch.object( - MCPRequestHandler, "_get_allowed_mcp_servers_for_agent" - ) as mock_agent: + with patch.object(MCPRequestHandler, "_get_allowed_mcp_servers_for_key") as mock_key: + with patch.object(MCPRequestHandler, "_get_allowed_mcp_servers_for_team") as mock_team: + with patch.object(MCPRequestHandler, "_get_allowed_mcp_servers_for_agent") as mock_agent: mock_key.return_value = ["server_1", "server_2"] mock_team.return_value = [] mock_agent.return_value = ["server_1"] - result = await MCPRequestHandler.get_allowed_mcp_servers( - user_api_key_auth=user_api_key_auth - ) + result = await MCPRequestHandler.get_allowed_mcp_servers(user_api_key_auth=user_api_key_auth) assert sorted(result) == ["server_1"] mock_agent.assert_called_once_with(user_api_key_auth) @@ -3644,21 +3578,13 @@ class TestAgentMCPPermissions: user_id="test-user", agent_id="agent-456", ) - with patch.object( - MCPRequestHandler, "_get_allowed_mcp_servers_for_key" - ) as mock_key: - with patch.object( - MCPRequestHandler, "_get_allowed_mcp_servers_for_team" - ) as mock_team: - with patch.object( - MCPRequestHandler, "_get_allowed_mcp_servers_for_agent" - ) as mock_agent: + with patch.object(MCPRequestHandler, "_get_allowed_mcp_servers_for_key") as mock_key: + with patch.object(MCPRequestHandler, "_get_allowed_mcp_servers_for_team") as mock_team: + with patch.object(MCPRequestHandler, "_get_allowed_mcp_servers_for_agent") as mock_agent: mock_key.return_value = ["server_1", "server_2"] mock_team.return_value = [] mock_agent.return_value = [] # no agent-level restriction - result = await MCPRequestHandler.get_allowed_mcp_servers( - user_api_key_auth=user_api_key_auth - ) + result = await MCPRequestHandler.get_allowed_mcp_servers(user_api_key_auth=user_api_key_auth) assert sorted(result) == ["server_1", "server_2"] mock_agent.assert_called_once_with(user_api_key_auth) @@ -3669,21 +3595,13 @@ class TestAgentMCPPermissions: user_id="test-user", agent_id="agent-789", ) - with patch.object( - MCPRequestHandler, "_get_allowed_mcp_servers_for_key" - ) as mock_key: - with patch.object( - MCPRequestHandler, "_get_allowed_mcp_servers_for_team" - ) as mock_team: - with patch.object( - MCPRequestHandler, "_get_allowed_mcp_servers_for_agent" - ) as mock_agent: + with patch.object(MCPRequestHandler, "_get_allowed_mcp_servers_for_key") as mock_key: + with patch.object(MCPRequestHandler, "_get_allowed_mcp_servers_for_team") as mock_team: + with patch.object(MCPRequestHandler, "_get_allowed_mcp_servers_for_agent") as mock_agent: mock_key.return_value = ["server_1", "server_2"] mock_team.return_value = [] mock_agent.return_value = ["server_2", "server_3"] - result = await MCPRequestHandler.get_allowed_mcp_servers( - user_api_key_auth=user_api_key_auth - ) + result = await MCPRequestHandler.get_allowed_mcp_servers(user_api_key_auth=user_api_key_auth) assert sorted(result) == ["server_2"] async def test_get_allowed_tools_for_server_agent_intersection(self): @@ -3696,9 +3614,7 @@ class TestAgentMCPPermissions: key_perm = MagicMock() key_perm.mcp_tool_permissions = {"server_1": ["tool_a", "tool_b"]} team_perm = None - with patch.object( - MCPRequestHandler, "_get_key_object_permission", return_value=key_perm - ): + with patch.object(MCPRequestHandler, "_get_key_object_permission", return_value=key_perm): with patch.object( MCPRequestHandler, "_get_team_object_permission", @@ -3730,9 +3646,7 @@ class TestAgentMCPPermissions: ) key_perm = MagicMock() key_perm.mcp_tool_permissions = {"server_1": ["tool_a", "tool_b"]} - with patch.object( - MCPRequestHandler, "_get_key_object_permission", return_value=key_perm - ): + with patch.object(MCPRequestHandler, "_get_key_object_permission", return_value=key_perm): with patch.object( MCPRequestHandler, "_get_team_object_permission", @@ -3762,9 +3676,7 @@ class TestAgentMCPPermissions: agent_row = MagicMock() agent_row.object_permission_id = "perm-xyz" prisma_client = MagicMock() - prisma_client.db.litellm_agentstable.find_unique = AsyncMock( - return_value=agent_row - ) + prisma_client.db.litellm_agentstable.find_unique = AsyncMock(return_value=agent_row) user_api_key_auth = UserAPIKeyAuth( api_key="test-key", user_id="test-user", @@ -3782,9 +3694,7 @@ class TestAgentMCPPermissions: return_value=expected_perm, ) as mock_get_perm, ): - result = await MCPRequestHandler._get_agent_object_permission( - user_api_key_auth - ) + result = await MCPRequestHandler._get_agent_object_permission(user_api_key_auth) assert result is expected_perm mock_get_perm.assert_awaited_once() assert mock_get_perm.await_args.kwargs["object_permission_id"] == "perm-xyz" @@ -3804,9 +3714,7 @@ class TestAgentMCPPermissions: agent_row = MagicMock() agent_row.object_permission_id = None prisma_client = MagicMock() - prisma_client.db.litellm_agentstable.find_unique = AsyncMock( - return_value=agent_row - ) + prisma_client.db.litellm_agentstable.find_unique = AsyncMock(return_value=agent_row) user_api_key_auth = UserAPIKeyAuth( api_key="test-key", user_id="test-user", @@ -3822,14 +3730,8 @@ class TestAgentMCPPermissions: new_callable=AsyncMock, ) as mock_get_perm, ): - assert ( - await MCPRequestHandler._get_agent_object_permission(user_api_key_auth) - is None - ) - assert ( - await MCPRequestHandler._get_agent_object_permission(user_api_key_auth) - is None - ) + assert await MCPRequestHandler._get_agent_object_permission(user_api_key_auth) is None + assert await MCPRequestHandler._get_agent_object_permission(user_api_key_auth) is None mock_get_perm.assert_not_awaited() prisma_client.db.litellm_agentstable.find_unique.assert_awaited_once() @@ -3870,9 +3772,7 @@ async def test_tool_permission_servers_included_in_allowed_servers(): ) with ( - patch.object( - MCPRequestHandler, "_get_key_object_permission", return_value=perm - ), + patch.object(MCPRequestHandler, "_get_key_object_permission", return_value=perm), patch.object( MCPRequestHandler, "_get_mcp_servers_from_access_groups", @@ -4105,9 +4005,7 @@ class TestOrgMCPPermissions: org_perm.mcp_tool_permissions = {"server_1": ["tool_a", "tool_b"]} with ( - patch.object( - MCPRequestHandler, "_get_key_object_permission", return_value=key_perm - ), + patch.object(MCPRequestHandler, "_get_key_object_permission", return_value=key_perm), patch.object( MCPRequestHandler, "_get_team_object_permission", @@ -4137,9 +4035,7 @@ class TestOrgMCPPermissions: org_perm.mcp_tool_permissions = {} with ( - patch.object( - MCPRequestHandler, "_get_key_object_permission", return_value=key_perm - ), + patch.object(MCPRequestHandler, "_get_key_object_permission", return_value=key_perm), patch.object( MCPRequestHandler, "_get_team_object_permission", @@ -4231,9 +4127,7 @@ async def test_mcp_key_access_group_extras_when_team_authorized(): ] _start_patches(patches) try: - result = await MCPRequestHandler._get_key_access_group_mcp_server_extras( - valid_token - ) + result = await MCPRequestHandler._get_key_access_group_mcp_server_extras(valid_token) assert result == ["srv-stripe"] finally: _stop_patches(patches) @@ -4270,9 +4164,7 @@ async def test_mcp_key_access_group_extras_when_key_directly_authorized(): ] _start_patches(patches) try: - result = await MCPRequestHandler._get_key_access_group_mcp_server_extras( - valid_token - ) + result = await MCPRequestHandler._get_key_access_group_mcp_server_extras(valid_token) assert result == ["srv-stripe"] finally: _stop_patches(patches) @@ -4286,9 +4178,7 @@ async def test_mcp_key_access_group_extras_when_key_has_no_groups(): access_group_ids=[], team_id="team-a", ) - result = await MCPRequestHandler._get_key_access_group_mcp_server_extras( - valid_token - ) + result = await MCPRequestHandler._get_key_access_group_mcp_server_extras(valid_token) assert result == [] @@ -4315,9 +4205,7 @@ async def test_mcp_key_access_group_extras_when_group_has_no_servers(): ] _start_patches(patches) try: - result = await MCPRequestHandler._get_key_access_group_mcp_server_extras( - valid_token - ) + result = await MCPRequestHandler._get_key_access_group_mcp_server_extras(valid_token) assert result == [] finally: _stop_patches(patches) @@ -4351,9 +4239,7 @@ async def test_mcp_key_access_group_extras_granted_even_when_group_authorizes_ne ] _start_patches(patches) try: - result = await MCPRequestHandler._get_key_access_group_mcp_server_extras( - valid_token - ) + result = await MCPRequestHandler._get_key_access_group_mcp_server_extras(valid_token) assert result == ["srv-finance-only"] finally: _stop_patches(patches) @@ -4376,9 +4262,7 @@ async def test_mcp_key_access_group_extras_when_get_access_object_raises(): ] _start_patches(patches) try: - result = await MCPRequestHandler._get_key_access_group_mcp_server_extras( - valid_token - ) + result = await MCPRequestHandler._get_key_access_group_mcp_server_extras(valid_token) assert result == [] finally: _stop_patches(patches) diff --git a/tests/test_litellm/proxy/_experimental/mcp_server/outbound_credentials/test_adapter.py b/tests/test_litellm/proxy/_experimental/mcp_server/outbound_credentials/test_adapter.py index fbe768e07c6..17960e917a4 100644 --- a/tests/test_litellm/proxy/_experimental/mcp_server/outbound_credentials/test_adapter.py +++ b/tests/test_litellm/proxy/_experimental/mcp_server/outbound_credentials/test_adapter.py @@ -23,6 +23,7 @@ from litellm.proxy._experimental.mcp_server.outbound_credentials.types import ( AuthorizationCodeConfig, CredError, NoneConfig, + PassthroughConfig, SharedKey, TokenExchangeConfig, ) @@ -234,6 +235,12 @@ def test_token_exchange_empty_subject_token_type_normalizes_to_default(): assert spec.config.subject_token_type == "urn:ietf:params:oauth:token-type:access_token" +@pytest.mark.parametrize("auth_type", [MCPAuth.true_passthrough, MCPAuth.oauth_delegate]) +def test_client_forwarded_modes_map_to_passthrough_config(auth_type): + spec = to_server_spec(_server(auth_type=auth_type)) + assert spec is not None and isinstance(spec.config, PassthroughConfig) + + @pytest.mark.parametrize( "server", [ diff --git a/tests/test_litellm/proxy/_experimental/mcp_server/outbound_credentials/test_resolver.py b/tests/test_litellm/proxy/_experimental/mcp_server/outbound_credentials/test_resolver.py index f8e45b38b49..64226eea821 100644 --- a/tests/test_litellm/proxy/_experimental/mcp_server/outbound_credentials/test_resolver.py +++ b/tests/test_litellm/proxy/_experimental/mcp_server/outbound_credentials/test_resolver.py @@ -1,9 +1,9 @@ """Tests for the resolver dispatch: live arms produce auth, stubbed arms fail closed. -`none`, `api_key` (shared-key source), `authorization_code`, and `token_exchange` are implemented; -every other arm, plus the `api_key` BYOK source, returns a typed `not_implemented` error until its -mode lands. Parametrizing the stubs over one config each also guards reachability: a dropped `case` -would hit `assert_never` and raise instead of returning the stub. +`none`, `api_key` (shared-key source), `passthrough`, `authorization_code`, and `token_exchange` are +implemented; every other arm, plus the `api_key` BYOK source, returns a typed `not_implemented` error +until its mode lands. Parametrizing the stubs over one config each also guards reachability: a dropped +`case` would hit `assert_never` and raise instead of returning the stub. """ import httpx @@ -253,9 +253,24 @@ async def test_token_exchange_without_an_exchanger_fails_closed(): assert result.error.tag == "misconfigured" +@pytest.mark.asyncio +async def test_passthrough_forwards_the_inbound_token_verbatim(): + subject = Subject(tenant_id="", subject_id="", inbound_token=SecretStr("Bearer upstream-xyz")) + result = await UpstreamCredentialProvider().resolve_credentials(subject, _spec(PassthroughConfig())) + assert isinstance(result, Ok) + assert isinstance(result.ok, StaticHeaderAuth) + assert _emitted(result.ok)["Authorization"] == "Bearer upstream-xyz" + + +@pytest.mark.asyncio +async def test_passthrough_without_inbound_token_is_a_no_op(): + result = await UpstreamCredentialProvider().resolve_credentials(_SUBJECT, _spec(PassthroughConfig())) + assert isinstance(result, Ok) + assert isinstance(result.ok, NoOpAuth) + + _STUBBED = [ ("api_key_byok", ApiKeyConfig(key_source=Byok())), - ("passthrough", PassthroughConfig()), ("client_credentials", ClientCredentialsConfig()), ("aws_sigv4", AwsSigV4Config(region="us-east-1")), ] diff --git a/tests/test_litellm/proxy/_experimental/mcp_server/test_db_credentials.py b/tests/test_litellm/proxy/_experimental/mcp_server/test_db_credentials.py index 8b8b8a363d5..7269774442b 100644 --- a/tests/test_litellm/proxy/_experimental/mcp_server/test_db_credentials.py +++ b/tests/test_litellm/proxy/_experimental/mcp_server/test_db_credentials.py @@ -11,6 +11,7 @@ keeps a plain-base64 fallback on read so existing rows continue to work. import base64 import json from datetime import datetime, timedelta, timezone +from types import SimpleNamespace from unittest.mock import AsyncMock, MagicMock import pytest @@ -63,6 +64,264 @@ def _legacy_row(payload: str): return row +def _identity_server(**overrides): + base = dict( + url="https://up.example.com/mcp", + auth_type="oauth2", + oauth2_flow="authorization_code", + authorization_url="https://idp.example.com/authorize", + token_url="https://idp.example.com/token", + registration_url="https://idp.example.com/register", + credentials={"client_id": "cid", "client_secret": "csec", "scopes": ["a"]}, + server_name="srv", + description="d", + ) + base.update(overrides) + return SimpleNamespace(**base) + + +@pytest.mark.parametrize( + "overrides", + [ + {"url": "https://other.example.com/mcp"}, + {"spec_path": "https://up.example.com/openapi.json"}, + {"auth_type": "oauth_delegate"}, + {"oauth2_flow": "client_credentials"}, + {"authorization_url": "https://other.example.com/authorize"}, + {"token_url": "https://other.example.com/token"}, + {"registration_url": "https://other.example.com/register"}, + {"credentials": {"client_id": "new", "client_secret": "csec", "scopes": ["a"]}}, + {"credentials": {"client_id": "cid", "client_secret": "rotated", "scopes": ["a"]}}, + {"credentials": {"client_id": "cid", "client_secret": "csec", "scopes": ["b"]}}, + ], +) +def test_mcp_oauth_token_identity_changes_on_mint_relevant_fields(overrides): + from litellm.proxy._experimental.mcp_server.db import mcp_oauth_token_identity + + assert mcp_oauth_token_identity(_identity_server()) != mcp_oauth_token_identity(_identity_server(**overrides)) + + +@pytest.mark.parametrize( + "overrides", + [ + {"server_name": "renamed"}, + {"description": "changed"}, + ], +) +def test_mcp_oauth_token_identity_stable_on_non_mint_fields(overrides): + from litellm.proxy._experimental.mcp_server.db import mcp_oauth_token_identity + + assert mcp_oauth_token_identity(_identity_server()) == mcp_oauth_token_identity(_identity_server(**overrides)) + + +def _encrypted_creds_json(client_id: str = "cid", client_secret: str = "csec") -> str: + from litellm.proxy._experimental.mcp_server.db import encrypt_credentials + + encrypted = encrypt_credentials( + credentials={"client_id": client_id, "client_secret": client_secret, "scopes": ["a"]}, + encryption_key=None, + ) + return json.dumps(encrypted) + + +def test_mcp_oauth_token_identity_stable_across_reencryption(): + """Stored client_id/client_secret are NaCl-encrypted with a fresh nonce on every write, so two + saves of the SAME plaintext produce different ciphertext. The identity must compare decrypted + values; comparing ciphertext would flag every routine save as a mint-relevant change and purge + per-user tokens that are still valid.""" + from litellm.proxy._experimental.mcp_server.db import mcp_oauth_token_identity + + first = _encrypted_creds_json() + second = _encrypted_creds_json() + assert first != second + + assert mcp_oauth_token_identity(_identity_server(credentials=first)) == mcp_oauth_token_identity( + _identity_server(credentials=second) + ) + + +def test_mcp_oauth_token_identity_detects_change_under_encryption(): + from litellm.proxy._experimental.mcp_server.db import mcp_oauth_token_identity + + unchanged = _identity_server(credentials=_encrypted_creds_json()) + changed = _identity_server(credentials=_encrypted_creds_json(client_id="other")) + assert mcp_oauth_token_identity(unchanged) != mcp_oauth_token_identity(changed) + + +def _oauth_row(user_id: str, server_id: str = "srv-1"): + """A stored per-user OAuth token row (payload tagged type=oauth2, legacy plain-base64 encoding).""" + row = _legacy_row(json.dumps({"type": "oauth2", "access_token": "tok-" + user_id})) + row.user_id = user_id + row.server_id = server_id + return row + + +def _byok_row(user_id: str, server_id: str = "srv-1"): + """A stored BYOK API key row: the same column, but the payload is a plain string, not OAuth JSON.""" + row = _legacy_row("sk-byok-" + user_id) + row.user_id = user_id + row.server_id = server_id + return row + + +@pytest.mark.asyncio +async def test_purge_user_oauth_credentials_for_server_invalidates_each_user(): + """The purge must route each (user, server) row through the invalidator exactly once.""" + from litellm.proxy._experimental.mcp_server.db import purge_user_oauth_credentials_for_server + + prisma = MagicMock() + prisma.db.litellm_mcpusercredentials.find_many = AsyncMock(return_value=[_oauth_row("alice"), _oauth_row("bob")]) + prisma.db.litellm_mcpusercredentials.delete_many = AsyncMock(return_value=2) + + invalidations = [] + + async def record_invalidation(user_id: str, server_id: str) -> None: + invalidations.append((user_id, server_id)) + + purged = await purge_user_oauth_credentials_for_server(prisma, "srv-1", invalidate_token_cache=record_invalidation) + + assert purged == 2 + prisma.db.litellm_mcpusercredentials.delete_many.assert_awaited_once_with( + where={"server_id": "srv-1", "user_id": {"in": ["alice", "bob"]}} + ) + assert set(invalidations) == {("alice", "srv-1"), ("bob", "srv-1")} + + +@pytest.mark.asyncio +async def test_purge_user_oauth_credentials_for_server_spares_byok_rows(): + """Regression: the purge used to delete_many on server_id alone, wiping BYOK API keys that share + the LiteLLM_MCPUserCredentials table. Only rows holding an OAuth2 payload may be deleted (one + batched query filtered to their user_ids), and only their users' token caches invalidated.""" + from litellm.proxy._experimental.mcp_server.db import purge_user_oauth_credentials_for_server + + prisma = MagicMock() + prisma.db.litellm_mcpusercredentials.find_many = AsyncMock(return_value=[_byok_row("carol"), _oauth_row("alice")]) + prisma.db.litellm_mcpusercredentials.delete_many = AsyncMock(return_value=1) + + invalidations = [] + + async def record_invalidation(user_id: str, server_id: str) -> None: + invalidations.append((user_id, server_id)) + + purged = await purge_user_oauth_credentials_for_server(prisma, "srv-1", invalidate_token_cache=record_invalidation) + + assert purged == 1 + prisma.db.litellm_mcpusercredentials.delete_many.assert_awaited_once_with( + where={"server_id": "srv-1", "user_id": {"in": ["alice"]}} + ) + assert invalidations == [("alice", "srv-1")] + + +@pytest.mark.asyncio +async def test_purge_user_oauth_credentials_for_server_all_byok_is_noop(): + """An api_key (BYOK-only) server whose identity tuple changes (e.g. its url) must purge nothing.""" + from litellm.proxy._experimental.mcp_server.db import purge_user_oauth_credentials_for_server + + prisma = MagicMock() + prisma.db.litellm_mcpusercredentials.find_many = AsyncMock(return_value=[_byok_row("carol"), _byok_row("dave")]) + prisma.db.litellm_mcpusercredentials.delete_many = AsyncMock() + + purged = await purge_user_oauth_credentials_for_server(prisma, "srv-1") + + assert purged == 0 + prisma.db.litellm_mcpusercredentials.delete_many.assert_not_awaited() + + +@pytest.mark.asyncio +async def test_purge_user_oauth_credentials_for_server_defaults_to_manager_invalidator(monkeypatch): + """When no invalidator is injected, the purge must resolve to the manager's shared + invalidate_user_oauth_token_cache, the single point covering both the legacy per-user token cache + and the v2 per-user OAuth token store; a wrong or no-op default silently leaves every cache + serving tokens minted for the superseded config.""" + from litellm.proxy._experimental.mcp_server import mcp_server_manager + from litellm.proxy._experimental.mcp_server.db import purge_user_oauth_credentials_for_server + + prisma = MagicMock() + prisma.db.litellm_mcpusercredentials.find_many = AsyncMock(return_value=[_oauth_row("alice")]) + prisma.db.litellm_mcpusercredentials.delete_many = AsyncMock(return_value=1) + + shared_invalidator = AsyncMock() + monkeypatch.setattr( + mcp_server_manager.global_mcp_server_manager, + "invalidate_user_oauth_token_cache", + shared_invalidator, + ) + + purged = await purge_user_oauth_credentials_for_server(prisma, "srv-1") + + assert purged == 1 + shared_invalidator.assert_awaited_once_with("alice", "srv-1") + + +@pytest.mark.asyncio +async def test_purge_user_oauth_credentials_for_server_logs_raced_rows(monkeypatch): + from litellm.proxy._experimental.mcp_server import db as db_module + from litellm.proxy._experimental.mcp_server.db import purge_user_oauth_credentials_for_server + + prisma = MagicMock() + prisma.db.litellm_mcpusercredentials.find_many = AsyncMock(return_value=[_oauth_row("alice")]) + prisma.db.litellm_mcpusercredentials.delete_many = AsyncMock(return_value=0) + warning = MagicMock() + monkeypatch.setattr(db_module.verbose_proxy_logger, "warning", warning) + + purged = await purge_user_oauth_credentials_for_server(prisma, "srv-1", invalidate_token_cache=AsyncMock()) + + assert purged == 0 + warning.assert_called_once() + + +@pytest.mark.asyncio +async def test_delete_mcp_server_invalidates_cached_tokens_for_enumerated_users(): + """Deleting a server must invalidate each enumerated user's cached per-user token: the caches are + keyed by (user_id, server_id), so a re-created server reusing the same server_id would otherwise + serve tokens minted for the deleted server until TTL.""" + from litellm.proxy._experimental.mcp_server.db import delete_mcp_server + + prisma = MagicMock() + prisma.db.litellm_mcpservertable.delete = AsyncMock(return_value=MagicMock(server_id="srv-1")) + prisma.db.litellm_mcpusercredentials.find_many = AsyncMock(return_value=[_oauth_row("alice"), _byok_row("bob")]) + prisma.db.litellm_mcpusercredentials.delete_many = AsyncMock(return_value=2) + prisma.db.litellm_mcpuserenvvars.delete_many = AsyncMock(return_value=0) + + invalidations = [] + + async def record_invalidation(user_id: str, server_id: str) -> None: + invalidations.append((user_id, server_id)) + + deleted = await delete_mcp_server(prisma, "srv-1", invalidate_token_cache=record_invalidation) + + assert deleted is not None + assert set(invalidations) == {("alice", "srv-1"), ("bob", "srv-1")} + + +@pytest.mark.asyncio +async def test_delete_mcp_server_returns_none_without_cleanup_when_server_missing(): + from litellm.proxy._experimental.mcp_server.db import delete_mcp_server + + prisma = MagicMock() + prisma.db.litellm_mcpservertable.delete = AsyncMock(return_value=None) + prisma.db.litellm_mcpusercredentials.find_many = AsyncMock() + + deleted = await delete_mcp_server(prisma, "srv-1", invalidate_token_cache=AsyncMock()) + + assert deleted is None + prisma.db.litellm_mcpusercredentials.find_many.assert_not_awaited() + + +@pytest.mark.asyncio +async def test_purge_user_oauth_credentials_for_server_noop_when_empty(): + from litellm.proxy._experimental.mcp_server.db import purge_user_oauth_credentials_for_server + + prisma = MagicMock() + prisma.db.litellm_mcpusercredentials.find_many = AsyncMock(return_value=[]) + prisma.db.litellm_mcpusercredentials.delete_many = AsyncMock() + + purged = await purge_user_oauth_credentials_for_server(prisma, "srv-1") + + assert purged == 0 + prisma.db.litellm_mcpusercredentials.delete_many.assert_not_awaited() + + def _stored_value(prisma) -> str: """Pull the credential_b64 value passed to the most recent upsert call.""" call = prisma.db.litellm_mcpusercredentials.upsert.call_args @@ -136,9 +395,7 @@ async def test_store_user_oauth_credential_does_not_persist_plaintext(): access_token = "ya29.a0AfH6SMBverysecretaccesstoken" prisma = _make_prisma_with_existing(row=None) - await store_user_oauth_credential( - prisma, "alice", "srv-1", access_token, refresh_token="rfr-xyz" - ) + await store_user_oauth_credential(prisma, "alice", "srv-1", access_token, refresh_token="rfr-xyz") stored = _stored_value(prisma) try: @@ -221,9 +478,7 @@ async def test_byok_guard_rejects_overwriting_encrypted_byok(): encrypted_row = MagicMock() encrypted_row.credential_b64 = _stored_value(prisma) - prisma.db.litellm_mcpusercredentials.find_unique = AsyncMock( - return_value=encrypted_row - ) + prisma.db.litellm_mcpusercredentials.find_unique = AsyncMock(return_value=encrypted_row) with pytest.raises(ValueError, match="could not be verified as an OAuth2"): await store_user_oauth_credential(prisma, "alice", "srv-1", "tok") @@ -265,18 +520,14 @@ async def test_list_oauth_credentials_filters_byok_and_returns_payloads(): "connected_at": "2024-01-01T00:00:00Z", } legacy_row = MagicMock() - legacy_row.credential_b64 = base64.urlsafe_b64encode( - json.dumps(legacy_payload).encode() - ).decode() + legacy_row.credential_b64 = base64.urlsafe_b64encode(json.dumps(legacy_payload).encode()).decode() legacy_row.server_id = "srv-legacy" byok_row = MagicMock() byok_row.credential_b64 = base64.urlsafe_b64encode(b"plain-byok-key").decode() byok_row.server_id = "srv-byok" - prisma.db.litellm_mcpusercredentials.find_many = AsyncMock( - return_value=[encrypted_row, legacy_row, byok_row] - ) + prisma.db.litellm_mcpusercredentials.find_many = AsyncMock(return_value=[encrypted_row, legacy_row, byok_row]) results = await list_user_oauth_credentials(prisma, "alice") @@ -326,9 +577,7 @@ async def test_rotate_re_encrypts_byok_with_new_key(monkeypatch): prisma.db.litellm_mcpusercredentials.update = AsyncMock() new_master_key = "rotated-salt-key-9999-9999-9999-9999" - await rotate_mcp_user_credentials_master_key( - prisma_client=prisma, new_master_key=new_master_key - ) + await rotate_mcp_user_credentials_master_key(prisma_client=prisma, new_master_key=new_master_key) update_call = prisma.db.litellm_mcpusercredentials.update.call_args new_stored = update_call.kwargs["data"]["credential_b64"] @@ -356,19 +605,13 @@ async def test_rotate_migrates_legacy_plaintext_rows(monkeypatch): legacy_row.user_id = "alice" legacy_row.server_id = "srv-legacy" legacy_row.credential_b64 = base64.urlsafe_b64encode(b"legacy-plain").decode() - prisma.db.litellm_mcpusercredentials.find_many = AsyncMock( - return_value=[legacy_row] - ) + prisma.db.litellm_mcpusercredentials.find_many = AsyncMock(return_value=[legacy_row]) prisma.db.litellm_mcpusercredentials.update = AsyncMock() new_key = "another-rotation-key-aaaa-bbbb-cccc-dddd" - await rotate_mcp_user_credentials_master_key( - prisma_client=prisma, new_master_key=new_key - ) + await rotate_mcp_user_credentials_master_key(prisma_client=prisma, new_master_key=new_key) - new_stored = prisma.db.litellm_mcpusercredentials.update.call_args.kwargs["data"][ - "credential_b64" - ] + new_stored = prisma.db.litellm_mcpusercredentials.update.call_args.kwargs["data"]["credential_b64"] monkeypatch.setenv("LITELLM_SALT_KEY", new_key) assert ( decrypt_value_helper( @@ -396,14 +639,10 @@ async def test_rotate_skips_undecodable_rows(): good_row.server_id = "srv-ok" good_row.credential_b64 = base64.urlsafe_b64encode(b"good-byok").decode() - prisma.db.litellm_mcpusercredentials.find_many = AsyncMock( - return_value=[bad_row, good_row] - ) + prisma.db.litellm_mcpusercredentials.find_many = AsyncMock(return_value=[bad_row, good_row]) prisma.db.litellm_mcpusercredentials.update = AsyncMock() - await rotate_mcp_user_credentials_master_key( - prisma_client=prisma, new_master_key="new-key-xxxx" - ) + await rotate_mcp_user_credentials_master_key(prisma_client=prisma, new_master_key="new-key-xxxx") # Only one update call — the good row. assert prisma.db.litellm_mcpusercredentials.update.call_count == 1 @@ -419,9 +658,7 @@ def _oauth_cred(access_token="at-live", refresh_token=None, expires_in_seconds=N if refresh_token is not None: cred["refresh_token"] = refresh_token if expires_in_seconds is not None: - cred["expires_at"] = ( - datetime.now(timezone.utc) + timedelta(seconds=expires_in_seconds) - ).isoformat() + cred["expires_at"] = (datetime.now(timezone.utc) + timedelta(seconds=expires_in_seconds)).isoformat() return cred @@ -438,12 +675,7 @@ def test_expiry_buffer_treats_soon_to_expire_as_expired(): cred = _oauth_cred(expires_in_seconds=30) assert is_oauth_credential_expired(cred, buffer_seconds=60) is True # A token comfortably beyond the buffer stays valid. - assert ( - is_oauth_credential_expired( - _oauth_cred(expires_in_seconds=600), buffer_seconds=60 - ) - is False - ) + assert is_oauth_credential_expired(_oauth_cred(expires_in_seconds=600), buffer_seconds=60) is False def test_expiry_past_is_expired_regardless_of_buffer(): @@ -465,9 +697,7 @@ async def test_resolve_returns_valid_token_without_refreshing(monkeypatch): refresh = AsyncMock() monkeypatch.setattr(db_mod, "refresh_user_oauth_token", refresh) - cred = _oauth_cred( - access_token="at-live", refresh_token="rt-1", expires_in_seconds=600 - ) + cred = _oauth_cred(access_token="at-live", refresh_token="rt-1", expires_in_seconds=600) result = await resolve_valid_user_oauth_token( user_id="alice", server=MagicMock(), cred=cred, prisma_client=MagicMock() ) @@ -483,15 +713,11 @@ async def test_resolve_refreshes_expired_token_with_refresh_token(monkeypatch): # new token rather than returning None (which left the UI tool list empty). import litellm.proxy._experimental.mcp_server.db as db_mod - refreshed = _oauth_cred( - access_token="at-fresh", refresh_token="rt-2", expires_in_seconds=3600 - ) + refreshed = _oauth_cred(access_token="at-fresh", refresh_token="rt-2", expires_in_seconds=3600) refresh = AsyncMock(return_value=refreshed) monkeypatch.setattr(db_mod, "refresh_user_oauth_token", refresh) - expired = _oauth_cred( - access_token="at-dead", refresh_token="rt-1", expires_in_seconds=-5 - ) + expired = _oauth_cred(access_token="at-dead", refresh_token="rt-1", expires_in_seconds=-5) result = await resolve_valid_user_oauth_token( user_id="alice", server=MagicMock(), cred=expired, prisma_client=MagicMock() ) @@ -510,9 +736,7 @@ async def test_resolve_refreshes_token_expiring_within_buffer(monkeypatch): refresh = AsyncMock(return_value=refreshed) monkeypatch.setattr(db_mod, "refresh_user_oauth_token", refresh) - soon = _oauth_cred( - access_token="at-soon", refresh_token="rt-1", expires_in_seconds=30 - ) + soon = _oauth_cred(access_token="at-soon", refresh_token="rt-1", expires_in_seconds=30) result = await resolve_valid_user_oauth_token( user_id="alice", server=MagicMock(), cred=soon, prisma_client=MagicMock() ) @@ -546,9 +770,7 @@ async def test_resolve_returns_none_when_refresh_fails(monkeypatch): refresh = AsyncMock(return_value=None) monkeypatch.setattr(db_mod, "refresh_user_oauth_token", refresh) - expired = _oauth_cred( - access_token="at-dead", refresh_token="rt-1", expires_in_seconds=-5 - ) + expired = _oauth_cred(access_token="at-dead", refresh_token="rt-1", expires_in_seconds=-5) result = await resolve_valid_user_oauth_token( user_id="alice", server=MagicMock(), cred=expired, prisma_client=MagicMock() ) @@ -565,9 +787,7 @@ async def test_resolve_returns_none_for_missing_credential(monkeypatch): monkeypatch.setattr(db_mod, "refresh_user_oauth_token", refresh) assert ( - await resolve_valid_user_oauth_token( - user_id="alice", server=MagicMock(), cred=None, prisma_client=MagicMock() - ) + await resolve_valid_user_oauth_token(user_id="alice", server=MagicMock(), cred=None, prisma_client=MagicMock()) is None ) assert ( @@ -601,19 +821,13 @@ async def test_rotate_user_env_vars_re_encrypts_with_new_key(monkeypatch): encrypted_old = encrypt_value_helper(json.dumps(values)) prisma = MagicMock() - prisma.db.litellm_mcpuserenvvars.find_many = AsyncMock( - return_value=[_env_var_row(encrypted_old)] - ) + prisma.db.litellm_mcpuserenvvars.find_many = AsyncMock(return_value=[_env_var_row(encrypted_old)]) prisma.db.litellm_mcpuserenvvars.update = AsyncMock() new_master_key = "rotated-env-key-1111-2222-3333-4444" - await rotate_mcp_user_env_vars_master_key( - prisma_client=prisma, new_master_key=new_master_key - ) + await rotate_mcp_user_env_vars_master_key(prisma_client=prisma, new_master_key=new_master_key) - new_stored = prisma.db.litellm_mcpuserenvvars.update.call_args.kwargs["data"][ - "values_b64" - ] + new_stored = prisma.db.litellm_mcpuserenvvars.update.call_args.kwargs["data"]["values_b64"] assert new_stored != encrypted_old, "rotation must produce different ciphertext" monkeypatch.setenv("LITELLM_SALT_KEY", new_master_key) @@ -630,18 +844,14 @@ async def test_rotate_user_env_vars_re_encrypts_with_new_key(monkeypatch): async def test_rotate_user_env_vars_skips_undecryptable_rows(): # A corrupt row must be skipped (not overwritten) so recoverable data is # preserved and one bad row does not abort the rest of the rotation. - good = _env_var_row( - encrypt_value_helper(json.dumps({"A": "1"})), server_id="srv-ok" - ) + good = _env_var_row(encrypt_value_helper(json.dumps({"A": "1"})), server_id="srv-ok") bad = _env_var_row("!!! not encrypted !!!", server_id="srv-corrupt") prisma = MagicMock() prisma.db.litellm_mcpuserenvvars.find_many = AsyncMock(return_value=[bad, good]) prisma.db.litellm_mcpuserenvvars.update = AsyncMock() - await rotate_mcp_user_env_vars_master_key( - prisma_client=prisma, new_master_key="new-key-xxxx" - ) + await rotate_mcp_user_env_vars_master_key(prisma_client=prisma, new_master_key="new-key-xxxx") assert prisma.db.litellm_mcpuserenvvars.update.call_count == 1 where = prisma.db.litellm_mcpuserenvvars.update.call_args.kwargs["where"] @@ -669,9 +879,7 @@ async def test_refresh_user_oauth_token_uses_client_secret_basic(monkeypatch): monkeypatch.setattr(db_mod, "get_async_httpx_client", lambda **kwargs: mock_client) monkeypatch.setattr(db_mod, "store_user_oauth_credential", AsyncMock()) - monkeypatch.setattr( - db_mod, "get_user_oauth_credential", AsyncMock(return_value={"access_token": "new-at"}) - ) + monkeypatch.setattr(db_mod, "get_user_oauth_credential", AsyncMock(return_value={"access_token": "new-at"})) result = await db_mod.refresh_user_oauth_token( prisma_client=MagicMock(), @@ -710,9 +918,7 @@ async def test_refresh_user_oauth_token_defaults_to_client_secret_post(monkeypat monkeypatch.setattr(db_mod, "get_async_httpx_client", lambda **kwargs: mock_client) monkeypatch.setattr(db_mod, "store_user_oauth_credential", AsyncMock()) - monkeypatch.setattr( - db_mod, "get_user_oauth_credential", AsyncMock(return_value={"access_token": "new-at"}) - ) + monkeypatch.setattr(db_mod, "get_user_oauth_credential", AsyncMock(return_value={"access_token": "new-at"})) await db_mod.refresh_user_oauth_token( prisma_client=MagicMock(), diff --git a/tests/test_litellm/proxy/_experimental/mcp_server/test_discoverable_endpoints.py b/tests/test_litellm/proxy/_experimental/mcp_server/test_discoverable_endpoints.py index c808b17678a..9de342eabd7 100644 --- a/tests/test_litellm/proxy/_experimental/mcp_server/test_discoverable_endpoints.py +++ b/tests/test_litellm/proxy/_experimental/mcp_server/test_discoverable_endpoints.py @@ -6,6 +6,8 @@ from unittest.mock import AsyncMock, MagicMock, patch import pytest from fastapi import HTTPException +from litellm.types.mcp import MCPAuth + # Fixture to mock IP address check for all MCP tests # This prevents tests from failing due to IP-based access control @@ -119,6 +121,61 @@ async def test_authorize_endpoint_includes_response_type(): assert "scope=read+write" in response.headers["location"] +@pytest.mark.asyncio +@pytest.mark.parametrize("auth_type_value", ["true_passthrough", "oauth_delegate"]) +async def test_authorize_endpoint_allows_client_forwarded_modes(auth_type_value): + """The browser-only Authorize relays the gateway authorize flow for the client-forwarded + token modes; the oauth2-only gate must let them through and redirect to the upstream IdP.""" + try: + from fastapi import Request + + from litellm.proxy._experimental.mcp_server.discoverable_endpoints import ( + authorize, + ) + from litellm.proxy._experimental.mcp_server.mcp_server_manager import ( + global_mcp_server_manager, + ) + from litellm.proxy._types import MCPTransport + from litellm.types.mcp import MCPAuth + from litellm.types.mcp_server.mcp_server_manager import MCPServer + except ImportError: + pytest.skip("MCP discoverable endpoints not available") + + global_mcp_server_manager.registry.clear() + + server = MCPServer( + server_id="test_cf_server", + name="test_cf", + server_name="test_cf", + alias="test_cf", + transport=MCPTransport.http, + auth_type=MCPAuth(auth_type_value), + # Discovery stamps these onto the in-memory registry entry at build time. + authorization_url="https://provider.com/oauth/authorize", + token_url="https://provider.com/oauth/token", + ) + global_mcp_server_manager.registry[server.server_id] = server + + mock_request = MagicMock(spec=Request) + mock_request.base_url = "https://litellm.example.com/" + mock_request.headers = {} + + with patch("litellm.proxy._experimental.mcp_server.discoverable_endpoints.encrypt_value_helper") as mock_encrypt: + mock_encrypt.return_value = "mocked_encrypted_state" + + response = await authorize( + request=mock_request, + client_id="dcr_client_id", + mcp_server_name="test_cf", + redirect_uri="http://127.0.0.1:60108/callback", + state="test_state", + ) + + assert response.status_code == 307 + assert "https://provider.com/oauth/authorize" in response.headers["location"] + assert "client_id=dcr_client_id" in response.headers["location"] + + @pytest.mark.asyncio async def test_authorize_endpoint_preserves_existing_query_params(): """Test that authorize endpoint merges OAuth params with existing query params in authorization_url""" @@ -3348,6 +3405,83 @@ async def test_token_exchange_passes_through_upstream_expires_in(): assert body["expires_in"] == 43200 +async def _exchange_persistence_attempted_for_auth_type(auth_type) -> bool: + """Run exchange_token_with_server for a server of ``auth_type`` and report whether it attempted + to persist the exchanged token server-side. The client-forwarded token modes must not persist: + their contract is that the upstream token stays browser-held, minted/stored/refreshed nowhere.""" + from fastapi import Request + + from litellm.proxy._experimental.mcp_server.discoverable_endpoints import ( + exchange_token_with_server, + ) + from litellm.proxy._types import MCPTransport + from litellm.types.mcp_server.mcp_server_manager import MCPServer + + server = MCPServer( + server_id="t", + name="t", + server_name="t", + alias="t", + transport=MCPTransport.http, + auth_type=auth_type, + client_id="cid", + client_secret="cs", + authorization_url="https://provider.com/oauth/authorize", + token_url="https://provider.com/oauth/token", + ) + mock_request = MagicMock(spec=Request) + mock_request.base_url = "https://litellm.example.com/" + mock_request.headers = {} + + fake_http_response = MagicMock() + fake_http_response.json.return_value = {"access_token": "tok", "refresh_token": "r", "token_type": "Bearer"} + fake_http_response.raise_for_status = MagicMock() + fake_http_client = MagicMock() + fake_http_client.post = AsyncMock(return_value=fake_http_response) + + with ( + patch( + "litellm.proxy._experimental.mcp_server.discoverable_endpoints.get_async_httpx_client", + return_value=fake_http_client, + ), + patch( + "litellm.proxy._experimental.mcp_server.discoverable_endpoints._extract_user_id_from_request", + new_callable=AsyncMock, + return_value="admin-user", + ), + patch( + "litellm.proxy._experimental.mcp_server.discoverable_endpoints._store_per_user_token_server_side", + new_callable=AsyncMock, + ) as mock_store, + ): + await exchange_token_with_server( + request=mock_request, + mcp_server=server, + grant_type="authorization_code", + code="c", + redirect_uri="http://127.0.0.1:3000/cb", + client_id="cid", + client_secret=None, + code_verifier=None, + ) + return mock_store.await_count > 0 + + +@pytest.mark.asyncio +@pytest.mark.parametrize("auth_type", [MCPAuth.true_passthrough, MCPAuth.oauth_delegate]) +async def test_token_exchange_does_not_persist_for_client_forwarded_modes(auth_type): + """The browser-only Authorize for true_passthrough / oauth_delegate must not write the upstream + token to the DB: these modes forward a browser-held token and persist nothing server-side.""" + assert await _exchange_persistence_attempted_for_auth_type(auth_type) is False + + +@pytest.mark.asyncio +async def test_token_exchange_persists_for_oauth2(): + """Guard the test's own discriminator: a genuine oauth2 (authorization_code) server DOES persist, + so the passthrough no-persist assertion above is meaningful and not vacuously true.""" + assert await _exchange_persistence_attempted_for_auth_type(MCPAuth.oauth2) is True + + # ------------------------------------------------------------------- # OBO (token_exchange) Protected Resource Metadata: discovery must name the # JWT-auth issuer the client SSOs with, not the gateway. @@ -4156,3 +4290,62 @@ async def test_store_per_user_token_server_side_skips_invalidate_when_db_write_f invalidate_mock.assert_not_awaited() cache_set_mock.assert_not_awaited() + + +@pytest.mark.asyncio +async def test_token_exchange_pairs_client_secret_with_server_client_id(): + """Re-auth regression: the register short-circuit hands the browser a placeholder + ``client_secret: "dummy"``, which the browser echoes back to /token. The server-side + persisted client_id wins the resolution, so the secret must come from the same (server) + source; pairing the persisted public PKCE client (no stored secret) with the caller's + placeholder makes the IdP reject the exchange with 401 on every re-auth.""" + from fastapi import Request + + from litellm.proxy._experimental.mcp_server.discoverable_endpoints import ( + exchange_token_with_server, + ) + from litellm.proxy._types import MCPTransport + from litellm.types.mcp import MCPAuth + from litellm.types.mcp_server.mcp_server_manager import MCPServer + + server = MCPServer( + server_id="srv-1", + name="srv-1", + server_name="srv-1", + alias="srv-1", + transport=MCPTransport.http, + auth_type=MCPAuth.oauth2, + client_id="persisted-client", + client_secret=None, + authorization_url="https://provider.example/oauth/authorize", + token_url="https://provider.example/oauth/token", + ) + + mock_request = MagicMock(spec=Request) + mock_request.base_url = "https://litellm.example.com/" + mock_request.headers = {} + + mock_response = MagicMock() + mock_response.raise_for_status = MagicMock() + mock_response.json.return_value = {"access_token": "at", "token_type": "Bearer"} + mock_async_client = MagicMock() + mock_async_client.post = AsyncMock(return_value=mock_response) + + with patch( + "litellm.proxy._experimental.mcp_server.discoverable_endpoints.get_async_httpx_client", + return_value=mock_async_client, + ): + await exchange_token_with_server( + request=mock_request, + mcp_server=server, + grant_type="authorization_code", + code="auth-code", + redirect_uri="https://litellm.example.com/ui/mcp/oauth/callback", + client_id="srv-1", + client_secret="dummy", + code_verifier="verifier", + ) + + sent = mock_async_client.post.call_args.kwargs["data"] + assert sent["client_id"] == "persisted-client" + assert "client_secret" not in sent diff --git a/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_oauth_passthrough.py b/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_oauth_passthrough.py index ad78609ee18..ef2a8318d7a 100644 --- a/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_oauth_passthrough.py +++ b/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_oauth_passthrough.py @@ -34,8 +34,7 @@ from litellm.types.mcp_server.mcp_server_manager import MCPServer def _mock_mcp_client_ip(): """Bypass IP-based access control in tests.""" with patch( - "litellm.proxy._experimental.mcp_server.discoverable_endpoints" - ".IPAddressUtils.get_mcp_client_ip", + "litellm.proxy._experimental.mcp_server.discoverable_endpoints.IPAddressUtils.get_mcp_client_ip", return_value=None, ): yield @@ -191,9 +190,7 @@ async def test_oauth_protected_resource_passthrough_proxies_upstream_metadata(): extra_headers=["Authorization"], oauth_passthrough=True, ) - global_mcp_server_manager.registry[passthrough_server.server_id] = ( - passthrough_server - ) + global_mcp_server_manager.registry[passthrough_server.server_id] = passthrough_server upstream_payload = { "resource": "https://upstream.example.com/mcp", @@ -207,18 +204,14 @@ async def test_oauth_protected_resource_passthrough_proxies_upstream_metadata(): mock_client = MagicMock() mock_client.get = AsyncMock(return_value=mock_response) - with patch.object( - discoverable_endpoints, "get_async_httpx_client", return_value=mock_client - ): + with patch.object(discoverable_endpoints, "get_async_httpx_client", return_value=mock_client): result = await _build_oauth_protected_resource_response( request=_make_request(), mcp_server_name="sample_docs", use_standard_pattern=True, ) - assert result["authorization_servers"] == [ - "https://okta.example.com/oauth2/default" - ] + assert result["authorization_servers"] == ["https://okta.example.com/oauth2/default"] # resource is normalized to the gateway URL so bearers are sent back to us assert result["resource"].endswith("/mcp/sample_docs") assert result["scopes_supported"] == ["openid", "profile"] @@ -242,9 +235,7 @@ async def test_oauth_protected_resource_passthrough_cache_hit(): extra_headers=["Authorization"], oauth_passthrough=True, ) - global_mcp_server_manager.registry[passthrough_server.server_id] = ( - passthrough_server - ) + global_mcp_server_manager.registry[passthrough_server.server_id] = passthrough_server mock_response = MagicMock() mock_response.status_code = 200 @@ -254,9 +245,7 @@ async def test_oauth_protected_resource_passthrough_cache_hit(): mock_client = MagicMock() mock_client.get = AsyncMock(return_value=mock_response) - with patch.object( - discoverable_endpoints, "get_async_httpx_client", return_value=mock_client - ): + with patch.object(discoverable_endpoints, "get_async_httpx_client", return_value=mock_client): await _build_oauth_protected_resource_response( request=_make_request(), mcp_server_name="sample_docs", @@ -348,12 +337,8 @@ async def test_oauth_metadata_cache_expired_entry_is_refetched(): mock_client = MagicMock() mock_client.get = AsyncMock(return_value=mock_response) - with patch.object( - discoverable_endpoints, "get_async_httpx_client", return_value=mock_client - ): - result = await discoverable_endpoints.fetch_upstream_oauth_protected_resource( - passthrough_server - ) + with patch.object(discoverable_endpoints, "get_async_httpx_client", return_value=mock_client): + result = await discoverable_endpoints.fetch_upstream_oauth_protected_resource(passthrough_server) assert result == {"authorization_servers": ["https://fresh.example.com"]} assert mock_client.get.await_count == 1 @@ -377,16 +362,12 @@ async def test_oauth_protected_resource_passthrough_network_error_returns_502(): extra_headers=["Authorization"], oauth_passthrough=True, ) - global_mcp_server_manager.registry[passthrough_server.server_id] = ( - passthrough_server - ) + global_mcp_server_manager.registry[passthrough_server.server_id] = passthrough_server mock_client = MagicMock() mock_client.get = AsyncMock(side_effect=httpx.ConnectError("boom")) - with patch.object( - discoverable_endpoints, "get_async_httpx_client", return_value=mock_client - ): + with patch.object(discoverable_endpoints, "get_async_httpx_client", return_value=mock_client): with pytest.raises(HTTPException) as exc_info: await _build_oauth_protected_resource_response( request=_make_request(), @@ -414,16 +395,10 @@ async def test_fetch_upstream_metadata_returns_none_when_not_all_candidates_netw not_found_response = MagicMock() not_found_response.status_code = 404 mock_client = MagicMock() - mock_client.get = AsyncMock( - side_effect=[not_found_response, httpx.ConnectError("path fallback failed")] - ) + mock_client.get = AsyncMock(side_effect=[not_found_response, httpx.ConnectError("path fallback failed")]) - with patch.object( - discoverable_endpoints, "get_async_httpx_client", return_value=mock_client - ): - result = await discoverable_endpoints.fetch_upstream_oauth_protected_resource( - passthrough_server - ) + with patch.object(discoverable_endpoints, "get_async_httpx_client", return_value=mock_client): + result = await discoverable_endpoints.fetch_upstream_oauth_protected_resource(passthrough_server) assert result is None assert mock_client.get.await_count == 2 @@ -458,9 +433,7 @@ async def test_oauth_protected_resource_gateway_managed_unchanged(): mock_client = MagicMock() mock_client.get = AsyncMock() - with patch.object( - discoverable_endpoints, "get_async_httpx_client", return_value=mock_client - ): + with patch.object(discoverable_endpoints, "get_async_httpx_client", return_value=mock_client): result = await _build_oauth_protected_resource_response( request=_make_request(), mcp_server_name="keycloak_whoami", @@ -468,7 +441,97 @@ async def test_oauth_protected_resource_gateway_managed_unchanged(): ) mock_client.get.assert_not_awaited() - assert result["authorization_servers"] == [ - "https://gateway.example.com/keycloak_whoami" - ] + assert result["authorization_servers"] == ["https://gateway.example.com/keycloak_whoami"] assert result["scopes_supported"] == ["read"] + + +def _make_upstream_metadata_client() -> tuple[dict, MagicMock]: + upstream_payload = { + "resource": "https://upstream.example.com/mcp", + "authorization_servers": ["https://okta.example.com/oauth2/default"], + "scopes_supported": ["openid", "profile"], + "bearer_methods_supported": ["header"], + } + mock_response = MagicMock() + mock_response.status_code = 200 + mock_response.json.return_value = upstream_payload + mock_client = MagicMock() + mock_client.get = AsyncMock(return_value=mock_response) + return upstream_payload, mock_client + + +@pytest.mark.asyncio +async def test_oauth_protected_resource_oauth_delegate_returns_upstream_metadata_verbatim(): + """oauth_delegate discovery must return the upstream metadata verbatim, + resource included. The caller's token is forwarded to and validated by the + upstream, so its audience must be the upstream; rewriting resource to the + gateway would make a strict IdP refuse to mint it or the upstream reject it. + A regression that dropped oauth_delegate from the pass-through predicate would + fall through to the gateway-AS branch and advertise LiteLLM as the AS.""" + from litellm.proxy._experimental.mcp_server.mcp_server_manager import ( + global_mcp_server_manager, + ) + + global_mcp_server_manager.registry.clear() + delegate_server = MCPServer( + server_id="delegate-1", + name="sample_docs", + server_name="sample_docs", + alias="sample_docs", + url="https://upstream.example.com/mcp", + transport=MCPTransport.http, + auth_type=MCPAuth.oauth_delegate, + ) + global_mcp_server_manager.registry[delegate_server.server_id] = delegate_server + + upstream_payload, mock_client = _make_upstream_metadata_client() + try: + with patch.object(discoverable_endpoints, "get_async_httpx_client", return_value=mock_client): + result = await _build_oauth_protected_resource_response( + request=_make_request(), + mcp_server_name="sample_docs", + use_standard_pattern=True, + ) + + assert result == upstream_payload + assert result["authorization_servers"] == ["https://okta.example.com/oauth2/default"] + assert result["resource"] == "https://upstream.example.com/mcp" + finally: + global_mcp_server_manager.registry.clear() + + +@pytest.mark.asyncio +async def test_oauth_protected_resource_true_passthrough_returns_upstream_metadata_verbatim(): + """true_passthrough discovery must return the upstream metadata verbatim, + resource included, so the client treats the upstream as the resource and + authorizes directly against it. A regression that rewrote resource (the + gateway-proxied behavior) would break the transparent-proxy contract.""" + from litellm.proxy._experimental.mcp_server.mcp_server_manager import ( + global_mcp_server_manager, + ) + + global_mcp_server_manager.registry.clear() + true_passthrough_server = MCPServer( + server_id="tp-1", + name="sample_docs", + server_name="sample_docs", + alias="sample_docs", + url="https://upstream.example.com/mcp", + transport=MCPTransport.http, + auth_type=MCPAuth.true_passthrough, + ) + global_mcp_server_manager.registry[true_passthrough_server.server_id] = true_passthrough_server + + upstream_payload, mock_client = _make_upstream_metadata_client() + try: + with patch.object(discoverable_endpoints, "get_async_httpx_client", return_value=mock_client): + result = await _build_oauth_protected_resource_response( + request=_make_request(), + mcp_server_name="sample_docs", + use_standard_pattern=True, + ) + + assert result == upstream_payload + assert result["resource"] == "https://upstream.example.com/mcp" + finally: + global_mcp_server_manager.registry.clear() diff --git a/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_server.py b/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_server.py index b24457deabd..7d25e0ba493 100644 --- a/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_server.py +++ b/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_server.py @@ -8,8 +8,10 @@ from fastapi import HTTPException from mcp import ReadResourceResult, Resource from mcp.types import ( BlobResourceContents, + CallToolResult, Prompt, ResourceTemplate, + TextContent, TextResourceContents, ) @@ -69,9 +71,7 @@ async def test_mcp_server_tool_call_body_contains_request_data(): # Mock the add_litellm_data_to_request function to capture the data captured_data = {} - async def mock_add_litellm_data_to_request( - data, request, user_api_key_dict, proxy_config - ): + async def mock_add_litellm_data_to_request(data, request, user_api_key_dict, proxy_config): captured_data.update(data) # Simulate the proxy_server_request creation captured_data["proxy_server_request"] = { @@ -355,6 +355,122 @@ def test_prepare_mcp_server_headers_m2m_skips_authorization_from_raw_extra_heade assert extra_headers.get("X-Custom") == "trace" +def _client_forwarded_mode_server(server_id: str, auth_type) -> MCPServer: + return MCPServer( + server_id=server_id, + name=server_id, + transport=MCPTransport.http, + auth_type=auth_type, + ) + + +def _prepare_headers_in_scope(server: MCPServer, scope_servers): + from litellm.proxy._experimental.mcp_server.server import ( + _prepare_mcp_server_headers, + ) + + return _prepare_mcp_server_headers( + server=server, + mcp_server_auth_headers=None, + mcp_auth_header=None, + oauth2_headers={"Authorization": "Bearer upstream-token"}, + raw_headers={ + "x-litellm-api-key": "Bearer sk-litellm-key", + "authorization": "Bearer upstream-token", + }, + user_api_key_auth=UserAPIKeyAuth(api_key="sk-litellm-key"), + scope_servers=scope_servers, + ) + + +def test_prepare_mcp_server_headers_withholds_global_authorization_when_scope_fans_out(): + """One caller bearer must not be replayed against multiple upstreams (RFC 9700 + cross-resource replay): in a fan-out scope with a second Authorization-consuming + server, the client-forwarded modes get no global Authorization.""" + delegate = _client_forwarded_mode_server("od-fanout", MCPAuth.oauth_delegate) + second_consumer = _client_forwarded_mode_server("tp-fanout", MCPAuth.true_passthrough) + + _, extra_headers = _prepare_headers_in_scope(delegate, [delegate, second_consumer]) + + assert not extra_headers or "authorization" not in {k.lower() for k in extra_headers} + + +def test_prepare_mcp_server_headers_forwards_global_authorization_to_sole_consumer(): + """Non-consuming servers (static api_key) in scope do not make the forward ambiguous.""" + delegate = _client_forwarded_mode_server("od-sole", MCPAuth.oauth_delegate) + static_server = MCPServer( + server_id="static-api-key", + name="static-api-key", + transport=MCPTransport.http, + auth_type=MCPAuth.api_key, + authentication_token="static-key", + ) + + _, extra_headers = _prepare_headers_in_scope(delegate, [delegate, static_server]) + + assert extra_headers == {"Authorization": "Bearer upstream-token"} + + +def test_prepare_mcp_server_headers_scope_counts_legacy_delegate_as_consumer(): + """Legacy upstream-delegated oauth2 servers still receive the caller's Authorization on the + v1 path, so their presence in scope must suppress the new modes' forward too.""" + delegate = _client_forwarded_mode_server("od-vs-legacy", MCPAuth.oauth_delegate) + legacy_delegate = MCPServer( + server_id="legacy-delegate", + name="legacy-delegate", + transport=MCPTransport.http, + auth_type=MCPAuth.oauth2, + delegate_auth_to_upstream=True, + ) + + _, extra_headers = _prepare_headers_in_scope(delegate, [delegate, legacy_delegate]) + + assert not extra_headers or "authorization" not in {k.lower() for k in extra_headers} + + +def test_prepare_mcp_server_headers_fanout_withhold_survives_extra_headers_loop(): + """Regression: when fan-out withholds the request-wide Authorization from a client-forwarded + server, the later server.extra_headers copy loop must not re-add it from raw_headers even if + the server lists Authorization in extra_headers. Otherwise one bearer is replayed across every + consuming upstream in the scope (the exact cross-resource replay the withholding prevents).""" + delegate = MCPServer( + server_id="od-extra-hdr", + name="od-extra-hdr", + transport=MCPTransport.http, + auth_type=MCPAuth.oauth_delegate, + extra_headers=["Authorization"], + ) + second_consumer = _client_forwarded_mode_server("tp-peer", MCPAuth.true_passthrough) + + _, extra_headers = _prepare_headers_in_scope(delegate, [delegate, second_consumer]) + + assert not extra_headers or "authorization" not in {k.lower() for k in extra_headers} + + +def test_prepare_mcp_server_headers_sole_consumer_still_forwards_via_extra_headers(): + """Guard the fix does not over-withhold: with no second consumer in scope, a client-forwarded + server that lists Authorization in extra_headers still forwards the caller's bearer.""" + delegate = MCPServer( + server_id="od-extra-sole", + name="od-extra-sole", + transport=MCPTransport.http, + auth_type=MCPAuth.oauth_delegate, + extra_headers=["Authorization"], + ) + static_server = MCPServer( + server_id="static-peer", + name="static-peer", + transport=MCPTransport.http, + auth_type=MCPAuth.api_key, + authentication_token="static-key", + ) + + _, extra_headers = _prepare_headers_in_scope(delegate, [delegate, static_server]) + + assert extra_headers is not None + assert extra_headers.get("Authorization") == "Bearer upstream-token" + + @pytest.mark.asyncio async def test_call_tool_m2m_skips_authorization_headers(): """M2M call_tool must not forward caller Authorization in oauth2/raw headers.""" @@ -382,9 +498,7 @@ async def test_call_tool_m2m_skips_authorization_headers(): mock_client = MagicMock() mock_client.call_tool = AsyncMock(return_value=MagicMock()) - with patch.object( - manager, "_create_mcp_client", new=AsyncMock(return_value=mock_client) - ) as create_client_mock: + with patch.object(manager, "_create_mcp_client", new=AsyncMock(return_value=mock_client)) as create_client_mock: await manager._call_regular_mcp_tool( mcp_server=server, original_tool_name="echo", @@ -879,9 +993,7 @@ async def test_get_tools_from_mcp_servers_continues_when_one_server_fails(): # Mock global_mcp_server_manager mock_manager = MagicMock() - mock_manager.get_allowed_mcp_servers = AsyncMock( - return_value=["working_server", "failing_server"] - ) + mock_manager.get_allowed_mcp_servers = AsyncMock(return_value=["working_server", "failing_server"]) mock_manager.get_mcp_server_by_id = lambda server_id: ( working_server if server_id == "working_server" else failing_server ) @@ -942,9 +1054,7 @@ async def test_get_tools_from_mcp_servers_continues_when_one_server_fails(): ) # Verify success logging - mock_logger.info.assert_any_call( - "Successfully fetched 1 tools total from all MCP servers" - ) + mock_logger.info.assert_any_call("Successfully fetched 1 tools total from all MCP servers") @pytest.mark.asyncio @@ -985,9 +1095,7 @@ async def test_get_tools_from_mcp_servers_handles_all_servers_failing(): # Mock global_mcp_server_manager mock_manager = MagicMock() - mock_manager.get_allowed_mcp_servers = AsyncMock( - return_value=["failing_server1", "failing_server2"] - ) + mock_manager.get_allowed_mcp_servers = AsyncMock(return_value=["failing_server1", "failing_server2"]) mock_manager.get_mcp_server_by_id = lambda server_id: ( failing_server1 if server_id == "failing_server1" else failing_server2 ) @@ -1042,9 +1150,7 @@ async def test_get_tools_from_mcp_servers_handles_all_servers_failing(): ) # Verify total logging - mock_logger.info.assert_any_call( - "Successfully fetched 0 tools total from all MCP servers" - ) + mock_logger.info.assert_any_call("Successfully fetched 0 tools total from all MCP servers") @pytest.mark.asyncio @@ -1069,9 +1175,7 @@ async def test_mcp_server_tool_call_body_with_none_arguments(): # Mock the add_litellm_data_to_request function to capture the data captured_data = {} - async def mock_add_litellm_data_to_request( - data, request, user_api_key_dict, proxy_config - ): + async def mock_add_litellm_data_to_request(data, request, user_api_key_dict, proxy_config): captured_data.update(data) captured_data["proxy_server_request"] = { "url": str(request.url), @@ -1176,31 +1280,29 @@ async def test_concurrent_initialize_session_managers(): results = await asyncio.gather(*tasks, return_exceptions=True) # All tasks should complete successfully (no exceptions) - assert all( - result == "success" for result in results - ), f"Some tasks failed: {results}" + assert all(result == "success" for result in results), f"Some tasks failed: {results}" # Each session manager.run() should only be called once due to the lock - assert ( - mock_stateless_run.call_count == 1 - ), f"Expected 1 call to session_manager_stateless.run(), got {mock_stateless_run.call_count}" - assert ( - mock_stateful_run.call_count == 1 - ), f"Expected 1 call to session_manager_stateful.run(), got {mock_stateful_run.call_count}" - assert ( - mock_sse_run.call_count == 1 - ), f"Expected 1 call to sse_session_manager.run(), got {mock_sse_run.call_count}" + assert mock_stateless_run.call_count == 1, ( + f"Expected 1 call to session_manager_stateless.run(), got {mock_stateless_run.call_count}" + ) + assert mock_stateful_run.call_count == 1, ( + f"Expected 1 call to session_manager_stateful.run(), got {mock_stateful_run.call_count}" + ) + assert mock_sse_run.call_count == 1, ( + f"Expected 1 call to sse_session_manager.run(), got {mock_sse_run.call_count}" + ) # The context managers should only be entered once each - assert ( - mock_cm_stateless.__aenter__.call_count == 1 - ), f"Expected 1 call to stateless __aenter__, got {mock_cm_stateless.__aenter__.call_count}" - assert ( - mock_cm_stateful.__aenter__.call_count == 1 - ), f"Expected 1 call to stateful __aenter__, got {mock_cm_stateful.__aenter__.call_count}" - assert ( - mock_cm_sse.__aenter__.call_count == 1 - ), f"Expected 1 call to sse __aenter__, got {mock_cm_sse.__aenter__.call_count}" + assert mock_cm_stateless.__aenter__.call_count == 1, ( + f"Expected 1 call to stateless __aenter__, got {mock_cm_stateless.__aenter__.call_count}" + ) + assert mock_cm_stateful.__aenter__.call_count == 1, ( + f"Expected 1 call to stateful __aenter__, got {mock_cm_stateful.__aenter__.call_count}" + ) + assert mock_cm_sse.__aenter__.call_count == 1, ( + f"Expected 1 call to sse __aenter__, got {mock_cm_sse.__aenter__.call_count}" + ) # State should be properly set assert mcp_server._SESSION_MANAGERS_INITIALIZED is True @@ -1343,16 +1445,12 @@ async def test_mcp_routing_initialize_to_stateful_no_session_to_stateless(): # initialize → stateful init_body = b'{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05"}}' stateless_called, stateful_called = await make_request(init_body) - assert ( - stateful_called and not stateless_called - ), "initialize (no session) should route to stateful, not stateless" + assert stateful_called and not stateless_called, "initialize (no session) should route to stateful, not stateless" # tools/list → stateless tools_body = b'{"jsonrpc":"2.0","id":2,"method":"tools/list","params":{}}' stateless_called, stateful_called = await make_request(tools_body) - assert ( - stateless_called and not stateful_called - ), "tools/list (no session) should route to stateless, not stateful" + assert stateless_called and not stateful_called, "tools/list (no session) should route to stateless, not stateful" @pytest.mark.asyncio @@ -1437,9 +1535,9 @@ async def test_mcp_routing_chunked_initialize_to_stateful(): ): await handle_streamable_http_mcp(scope, receive, send) - assert ( - stateful_called and not stateless_called - ), "chunked initialize (no session) should route to stateful, not stateless" + assert stateful_called and not stateless_called, ( + "chunked initialize (no session) should route to stateful, not stateless" + ) @pytest.mark.asyncio @@ -1468,10 +1566,7 @@ async def test_mcp_routing_caps_body_peek_for_oversized_chunked_body(): messages = [ {"type": "http.request", "body": first_chunk, "more_body": True}, - *[ - {"type": "http.request", "body": chunk, "more_body": True} - for chunk in oversized_tail - ], + *[{"type": "http.request", "body": chunk, "more_body": True} for chunk in oversized_tail], {"type": "http.request", "body": b"", "more_body": False}, ] receive_calls = {"count": 0} @@ -1522,12 +1617,8 @@ async def test_mcp_routing_caps_body_peek_for_oversized_chunked_body(): "litellm.proxy._experimental.mcp_server.server._SESSION_MANAGERS_INITIALIZED", True, ), - patch.object( - session_manager_stateless, "handle_request", side_effect=stateless_handle - ), - patch.object( - session_manager_stateful, "handle_request", side_effect=stateful_handle - ), + patch.object(session_manager_stateless, "handle_request", side_effect=stateless_handle), + patch.object(session_manager_stateful, "handle_request", side_effect=stateful_handle), patch.object(session_manager_stateless, "_server_instances", {}), patch.object(session_manager_stateful, "_server_instances", {}), ): @@ -1578,9 +1669,7 @@ async def test_enforce_stateful_session_cap_evicts_oldest_idle_then_rejects(): patch.object(session_manager_stateful, "_server_instances", instances), patch.object(mcp_server, "_MAX_STATEFUL_SESSIONS_PER_OWNER", 3), patch.dict(mcp_server._stateful_session_owners, owners, clear=True), - patch.dict( - mcp_server._stateful_session_auth_context_last_seen, last_seen, clear=True - ), + patch.dict(mcp_server._stateful_session_auth_context_last_seen, last_seen, clear=True), patch.dict(mcp_server._stateful_session_auth_contexts, contexts, clear=True), patch.dict(mcp_server._stateful_session_active_request_counts, {}, clear=True), ): @@ -1593,9 +1682,7 @@ async def test_enforce_stateful_session_cap_evicts_oldest_idle_then_rejects(): # A different owner at the cap is unaffected by owner-A's sessions. terminated.clear() - allowed_other = await mcp_server._enforce_stateful_session_cap_for_owner( - "owner-B" - ) + allowed_other = await mcp_server._enforce_stateful_session_cap_for_owner("owner-B") assert allowed_other is True assert terminated == [] @@ -1614,9 +1701,7 @@ async def test_enforce_stateful_session_cap_evicts_oldest_idle_then_rejects(): {f"s{i}": float(i) for i in range(3)}, clear=True, ), - patch.dict( - mcp_server._stateful_session_active_request_counts, active, clear=True - ), + patch.dict(mcp_server._stateful_session_active_request_counts, active, clear=True), ): rejected = await mcp_server._enforce_stateful_session_cap_for_owner("owner-A") assert rejected is False @@ -1662,9 +1747,7 @@ async def test_mcp_routing_initialize_rejected_when_owner_at_session_cap(): (b"authorization", b"Bearer test-key"), ], } - receive = AsyncMock( - return_value={"type": "http.request", "body": init_body, "more_body": False} - ) + receive = AsyncMock(return_value={"type": "http.request", "body": init_body, "more_body": False}) send = AsyncMock() stateful_called = [] @@ -1685,9 +1768,7 @@ async def test_mcp_routing_initialize_rejected_when_owner_at_session_cap(): ), patch.object(mcp_server, "_owner_fingerprint_for", return_value="owner-X"), patch.object(mcp_server, "_MAX_STATEFUL_SESSIONS_PER_OWNER", cap), - patch.object( - session_manager_stateful, "handle_request", side_effect=stateful_handle - ), + patch.object(session_manager_stateful, "handle_request", side_effect=stateful_handle), patch.object(session_manager_stateful, "_server_instances", instances), patch.object(session_manager_stateless, "_server_instances", {}), patch.dict(mcp_server._stateful_session_owners, owners, clear=True), @@ -1696,18 +1777,14 @@ async def test_mcp_routing_initialize_rejected_when_owner_at_session_cap(): {f"s{i}": float(i) for i in range(cap)}, clear=True, ), - patch.dict( - mcp_server._stateful_session_active_request_counts, active, clear=True - ), + patch.dict(mcp_server._stateful_session_active_request_counts, active, clear=True), patch.dict(mcp_server._stateful_session_auth_contexts, contexts, clear=True), ): await handle_streamable_http_mcp(scope, receive, send) assert not stateful_called, "initialize at session cap must not reach the manager" start_messages = [ - call.args[0] - for call in send.call_args_list - if call.args and call.args[0].get("type") == "http.response.start" + call.args[0] for call in send.call_args_list if call.args and call.args[0].get("type") == "http.response.start" ] assert start_messages, "a response should have been sent" assert start_messages[0]["status"] == 429 @@ -1743,9 +1820,7 @@ async def test_stateful_mcp_requests_refresh_session_auth_context(): None, "1.1.1.1", ) - mcp_server._stateful_session_auth_contexts[session_id] = callback_context.run( - mcp_server.auth_context_var.get - ) + mcp_server._stateful_session_auth_contexts[session_id] = callback_context.run(mcp_server.auth_context_var.get) scope = { "type": "http", @@ -2006,24 +2081,14 @@ async def test_initialize_request_with_existing_session_tracks_new_session(): ) await mcp_server._purge_expired_stateful_session_auth_contexts(now=now) assert new_session_id in mcp_server._stateful_session_auth_contexts - assert ( - mcp_server._stateful_session_auth_contexts[new_session_id] - is not existing_auth_user - ) - assert ( - mcp_server._stateful_session_auth_contexts[new_session_id].mcp_auth_header - == "new-mcp-auth" - ) + assert mcp_server._stateful_session_auth_contexts[new_session_id] is not existing_auth_user + assert mcp_server._stateful_session_auth_contexts[new_session_id].mcp_auth_header == "new-mcp-auth" async def stateless_handle(s, r, se): - raise AssertionError( - "initialize request with session should use stateful manager" - ) + raise AssertionError("initialize request with session should use stateful manager") try: - mcp_server._stateful_session_auth_contexts[existing_session_id] = ( - existing_auth_user - ) + mcp_server._stateful_session_auth_contexts[existing_session_id] = existing_auth_user mcp_server._stateful_session_auth_context_last_seen[existing_session_id] = 1.0 mcp_server._stateful_session_owners[existing_session_id] = owner_fingerprint @@ -2065,10 +2130,7 @@ async def test_initialize_request_with_existing_session_tracks_new_session(): assert stateful_called assert new_session_id not in mcp_server._stateful_session_active_request_counts assert new_session_id in mcp_server._stateful_session_auth_contexts - assert ( - mcp_server._stateful_session_auth_contexts[existing_session_id] - is existing_auth_user - ) + assert mcp_server._stateful_session_auth_contexts[existing_session_id] is existing_auth_user assert existing_auth_user.mcp_auth_header == "old-mcp-auth" assert existing_auth_user.mcp_servers == ["old-server"] finally: @@ -2191,15 +2253,11 @@ async def test_stateful_mcp_cleanup_loop_survives_purge_errors(): except ImportError: pytest.skip("MCP server not available") - purge = AsyncMock( - side_effect=[RuntimeError("terminate failed"), asyncio.CancelledError()] - ) + purge = AsyncMock(side_effect=[RuntimeError("terminate failed"), asyncio.CancelledError()]) with ( patch.object(mcp_server.asyncio, "sleep", AsyncMock(return_value=None)), - patch.object( - mcp_server, "_purge_expired_stateful_session_auth_contexts", purge - ), + patch.object(mcp_server, "_purge_expired_stateful_session_auth_contexts", purge), ): with pytest.raises(asyncio.CancelledError): await mcp_server._cleanup_expired_stateful_session_auth_contexts() @@ -2289,9 +2347,7 @@ async def test_stateful_mcp_session_owner_mismatch_returns_403(): intruder_auth = UserAPIKeyAuth(api_key="intruder-key", user_id="intruder") mcp_server._stateful_session_auth_contexts[session_id] = MagicMock() - mcp_server._stateful_session_owners[session_id] = mcp_server._owner_fingerprint_for( - owner_auth - ) + mcp_server._stateful_session_owners[session_id] = mcp_server._owner_fingerprint_for(owner_auth) scope = { "type": "http", @@ -2341,9 +2397,7 @@ async def test_stateful_mcp_session_owner_mismatch_returns_403(): await handle_streamable_http_mcp(scope, receive, capture_send) handle_request_mock.assert_not_awaited() - statuses = [ - m["status"] for m in sent_messages if m.get("type") == "http.response.start" - ] + statuses = [m["status"] for m in sent_messages if m.get("type") == "http.response.start"] assert statuses == [403] mcp_server._stateful_session_auth_contexts.pop(session_id, None) @@ -2368,12 +2422,10 @@ async def test_stateful_mcp_session_serializes_concurrent_requests(): session_id = "serialized-session-1" owner_auth = UserAPIKeyAuth(api_key="owner-key", user_id="owner") - mcp_server._stateful_session_auth_contexts[session_id] = ( - mcp_server.MCPAuthenticatedUser(user_api_key_auth=owner_auth) - ) - mcp_server._stateful_session_owners[session_id] = mcp_server._owner_fingerprint_for( - owner_auth + mcp_server._stateful_session_auth_contexts[session_id] = mcp_server.MCPAuthenticatedUser( + user_api_key_auth=owner_auth ) + mcp_server._stateful_session_owners[session_id] = mcp_server._owner_fingerprint_for(owner_auth) inside = 0 max_inside = 0 @@ -2414,9 +2466,7 @@ async def test_stateful_mcp_session_serializes_concurrent_requests(): "litellm.proxy._experimental.mcp_server.server._SESSION_MANAGERS_INITIALIZED", True, ), - patch.object( - session_manager_stateful, "handle_request", side_effect=slow_handle - ), + patch.object(session_manager_stateful, "handle_request", side_effect=slow_handle), patch.object( session_manager_stateful, "_server_instances", @@ -2432,9 +2482,7 @@ async def test_stateful_mcp_session_serializes_concurrent_requests(): mcp_server._stateful_session_owners.pop(session_id, None) mcp_server._stateful_session_locks.pop(session_id, None) - assert ( - max_inside == 1 - ), "concurrent requests on same stateful session must be serialized" + assert max_inside == 1, "concurrent requests on same stateful session must be serialized" @pytest.mark.asyncio @@ -2486,9 +2534,7 @@ async def test_stateful_mcp_lock_does_not_leak_when_auth_context_missing(): "litellm.proxy._experimental.mcp_server.server._SESSION_MANAGERS_INITIALIZED", True, ), - patch.object( - session_manager_stateful, "handle_request", side_effect=handle - ), + patch.object(session_manager_stateful, "handle_request", side_effect=handle), patch.object( session_manager_stateful, "_server_instances", @@ -2498,9 +2544,9 @@ async def test_stateful_mcp_lock_does_not_leak_when_auth_context_missing(): assert session_id not in mcp_server._stateful_session_auth_contexts await handle_streamable_http_mcp(scope, receive, AsyncMock()) - assert ( - session_id not in mcp_server._stateful_session_locks - ), "lock entry must be cleaned up for untracked stateful session" + assert session_id not in mcp_server._stateful_session_locks, ( + "lock entry must be cleaned up for untracked stateful session" + ) finally: mcp_server._stateful_session_auth_contexts.pop(session_id, None) mcp_server._stateful_session_owners.pop(session_id, None) @@ -2526,12 +2572,10 @@ async def test_stateful_mcp_get_stream_does_not_block_post(): session_id = "stream-session-1" owner_auth = UserAPIKeyAuth(api_key="owner-key", user_id="owner") - mcp_server._stateful_session_auth_contexts[session_id] = ( - mcp_server.MCPAuthenticatedUser(user_api_key_auth=owner_auth) - ) - mcp_server._stateful_session_owners[session_id] = mcp_server._owner_fingerprint_for( - owner_auth + mcp_server._stateful_session_auth_contexts[session_id] = mcp_server.MCPAuthenticatedUser( + user_api_key_auth=owner_auth ) + mcp_server._stateful_session_owners[session_id] = mcp_server._owner_fingerprint_for(owner_auth) stream_release = asyncio.Event() post_finished = asyncio.Event() @@ -2569,9 +2613,7 @@ async def test_stateful_mcp_get_stream_does_not_block_post(): "litellm.proxy._experimental.mcp_server.server._SESSION_MANAGERS_INITIALIZED", True, ), - patch.object( - session_manager_stateful, "handle_request", side_effect=handle - ), + patch.object(session_manager_stateful, "handle_request", side_effect=handle), patch.object( session_manager_stateful, "_server_instances", @@ -2615,10 +2657,7 @@ def test_jsonrpc_text_has_top_level_method_ignores_nested_method(): assert _jsonrpc_text_has_top_level_method(reordered) is True # response whose result nests a "method" key (and arrays of them) - response = ( - '{"jsonrpc":"2.0","id":1,"result":{"toolResult":{"method":"GET"},' - '"steps":[{"method":"x"}]}}' - ) + response = '{"jsonrpc":"2.0","id":1,"result":{"toolResult":{"method":"GET"},"steps":[{"method":"x"}]}}' assert _jsonrpc_text_has_top_level_method(response) is False # truncated response: result value never closes, no top-level method seen @@ -2643,12 +2682,10 @@ async def test_truncated_jsonrpc_response_with_nested_method_skips_lock(): session_id = "nested-method-response-session" owner_auth = UserAPIKeyAuth(api_key="owner-key", user_id="owner") - mcp_server._stateful_session_auth_contexts[session_id] = ( - mcp_server.MCPAuthenticatedUser(user_api_key_auth=owner_auth) - ) - mcp_server._stateful_session_owners[session_id] = mcp_server._owner_fingerprint_for( - owner_auth + mcp_server._stateful_session_auth_contexts[session_id] = mcp_server.MCPAuthenticatedUser( + user_api_key_auth=owner_auth ) + mcp_server._stateful_session_owners[session_id] = mcp_server._owner_fingerprint_for(owner_auth) gate = asyncio.Event() request_in_handle = asyncio.Event() @@ -2685,8 +2722,7 @@ async def test_truncated_jsonrpc_response_with_nested_method_skips_lock(): # parsed, with a nested "method" key in the first bytes to trip a flat # substring heuristic. response_body = ( - '{"jsonrpc":"2.0","id":99,"result":{"toolResult":' - '{"method":"GET","payload":"' + ("x" * 5000) + '"}}}' + '{"jsonrpc":"2.0","id":99,"result":{"toolResult":{"method":"GET","payload":"' + ("x" * 5000) + '"}}}' ).encode() try: @@ -2700,9 +2736,7 @@ async def test_truncated_jsonrpc_response_with_nested_method_skips_lock(): "litellm.proxy._experimental.mcp_server.server._SESSION_MANAGERS_INITIALIZED", True, ), - patch.object( - session_manager_stateful, "handle_request", side_effect=handle - ), + patch.object(session_manager_stateful, "handle_request", side_effect=handle), patch.object( session_manager_stateful, "_server_instances", @@ -2774,13 +2808,9 @@ async def test_mcp_routing_with_conflicting_alias_and_group_name(): mock_get_tools_spy = AsyncMock(return_value=[]) # Mock the function that checks DB for an access group named "custom_solutions" - mock_db_lookup = AsyncMock( - return_value=[specific_server.server_id, other_server.server_id] - ) + mock_db_lookup = AsyncMock(return_value=[specific_server.server_id, other_server.server_id]) - mock_get_allowed = AsyncMock( - return_value=[specific_server.server_id, other_server.server_id] - ) + mock_get_allowed = AsyncMock(return_value=[specific_server.server_id, other_server.server_id]) with ( patch( @@ -2805,14 +2835,12 @@ async def test_mcp_routing_with_conflicting_alias_and_group_name(): ) # Get the list of actual server objects that the orchestrator tried to contact - called_servers = [ - call.kwargs["server"] for call in mock_get_tools_spy.call_args_list - ] + called_servers = [call.kwargs["server"] for call in mock_get_tools_spy.call_args_list] assert len(called_servers) == 1, "Should have resolved to exactly one server." - assert ( - called_servers[0].server_id == specific_server.server_id - ), "Should have contacted the specific server alias, not the group." + assert called_servers[0].server_id == specific_server.server_id, ( + "Should have contacted the specific server alias, not the group." + ) @pytest.mark.asyncio @@ -2926,9 +2954,7 @@ async def test_oauth2_caller_headers_not_forwarded_for_migrated_server(): ) # Verify that _create_mcp_client was called - assert ( - mock_create_client.call_count == 1 - ), "Expected _create_mcp_client to be called once" + assert mock_create_client.call_count == 1, "Expected _create_mcp_client to be called once" # Verify the server passed to _create_mcp_client is the OAuth2 server assert captured_client_args["server"].server_id == oauth2_server.server_id @@ -2938,9 +2964,9 @@ async def test_oauth2_caller_headers_not_forwarded_for_migrated_server(): # oauth2 Authorization upstream. The v2 resolver injects the stored per-user token, # so a caller-supplied bearer cannot override another user's stored credential. extra_headers = captured_client_args["extra_headers"] - assert extra_headers is None or "Authorization" not in { - k.lower() for k in extra_headers - }, f"Caller Authorization must not be forwarded, got {extra_headers}" + assert extra_headers is None or "Authorization" not in {k.lower() for k in extra_headers}, ( + f"Caller Authorization must not be forwarded, got {extra_headers}" + ) @pytest.mark.asyncio @@ -3049,12 +3075,8 @@ async def test_list_tools_multiple_servers_prefixed_names(): # Mock manager mock_manager = MagicMock() - mock_manager.get_allowed_mcp_servers = AsyncMock( - return_value=["server1", "server2"] - ) - mock_manager.get_mcp_server_by_id = lambda server_id: ( - server1 if server_id == "server1" else server2 - ) + mock_manager.get_allowed_mcp_servers = AsyncMock(return_value=["server1", "server2"]) + mock_manager.get_mcp_server_by_id = lambda server_id: server1 if server_id == "server1" else server2 # Mock filter_server_ids_by_ip to return server_ids unchanged (no IP filtering) mock_manager.filter_server_ids_by_ip_with_info = lambda server_ids, client_ip: ( server_ids, @@ -3653,9 +3675,7 @@ async def test_list_tools_strips_prefix_when_matching_permissions(): tool1.inputSchema = {} tool2 = MagicMock() - tool2.name = ( - "GITMCP-search_litellm_documentation" # Prefixed, not in allowed list - ) + tool2.name = "GITMCP-search_litellm_documentation" # Prefixed, not in allowed list tool2.description = "Search docs" tool2.inputSchema = {} @@ -3876,17 +3896,13 @@ class TestMCPServerManagerReload: db_row = _make_db_mcp_server("server-1", timestamp) mock_prisma = MagicMock() - mock_prisma.db.litellm_mcpservertable.find_many = AsyncMock( - return_value=[db_row] - ) + mock_prisma.db.litellm_mcpservertable.find_many = AsyncMock(return_value=[db_row]) with ( patch( "litellm.proxy.management_endpoints.mcp_management_endpoints.get_prisma_client_or_throw", return_value=mock_prisma, ), - patch.object( - manager, "build_mcp_server_from_table", AsyncMock() - ) as mock_build, + patch.object(manager, "build_mcp_server_from_table", AsyncMock()) as mock_build, ): await manager.reload_servers_from_database() @@ -3922,9 +3938,7 @@ class TestMCPServerManagerReload: ) mock_prisma = MagicMock() - mock_prisma.db.litellm_mcpservertable.find_many = AsyncMock( - return_value=[db_row] - ) + mock_prisma.db.litellm_mcpservertable.find_many = AsyncMock(return_value=[db_row]) with ( patch( "litellm.proxy.management_endpoints.mcp_management_endpoints.get_prisma_client_or_throw", @@ -4049,9 +4063,7 @@ class TestMCPServerManagerReload: raise RuntimeError("blocked address") mock_prisma = MagicMock() - mock_prisma.db.litellm_mcpservertable.find_many = AsyncMock( - return_value=[healthy_row, bad_openapi_row] - ) + mock_prisma.db.litellm_mcpservertable.find_many = AsyncMock(return_value=[healthy_row, bad_openapi_row]) with ( patch( "litellm.proxy.management_endpoints.mcp_management_endpoints.get_prisma_client_or_throw", @@ -4147,10 +4159,7 @@ async def test_call_mcp_tool_logs_failure_via_post_call_failure_hook(): ) proxy_logging_mock.post_call_failure_hook.assert_awaited_once() - assert ( - proxy_logging_mock.post_call_failure_hook.await_args.kwargs.get("route") - == "/mcp/call_tool" - ) + assert proxy_logging_mock.post_call_failure_hook.await_args.kwargs.get("route") == "/mcp/call_tool" @pytest.mark.asyncio @@ -4232,9 +4241,7 @@ async def test_get_tools_from_mcp_servers_logs_list_tools_to_spendlogs_when_enab assert tools == [tool_1] dummy_logging_obj.async_success_handler.assert_awaited_once() - assert dummy_logging_obj.async_success_handler.await_args.kwargs["result"] == [ - tool_1.model_dump(mode="json") - ] + assert dummy_logging_obj.async_success_handler.await_args.kwargs["result"] == [tool_1.model_dump(mode="json")] assert function_setup_kwargs["metadata"]["tags"] == ["team-a"] spend_meta = dummy_logging_obj.model_call_details["metadata"]["spend_logs_metadata"] @@ -4580,9 +4587,7 @@ async def test_get_tools_from_mcp_servers_injects_stored_oauth2_token(): oauth2_server.extra_headers = None # Simulate the DB returning a valid credential for this user+server - prefetched_creds = { - SERVER_ID: {"access_token": STORED_TOKEN, "server_id": SERVER_ID} - } + prefetched_creds = {SERVER_ID: {"access_token": STORED_TOKEN, "server_id": SERVER_ID}} tool_1 = MagicMock() tool_1.name = "atlassian_test-search" @@ -4689,16 +4694,12 @@ class TestMergeGatewayInitializeInstructions: global_mcp_server_manager, ) - global_mcp_server_manager._upstream_initialize_instructions_by_server_id[ - "s1" - ] = "upstream" + global_mcp_server_manager._upstream_initialize_instructions_by_server_id["s1"] = "upstream" try: s = _make_instruction_server(instructions="yaml wins") assert self._merge([s]) == "yaml wins" finally: - global_mcp_server_manager._upstream_initialize_instructions_by_server_id.pop( - "s1", None - ) + global_mcp_server_manager._upstream_initialize_instructions_by_server_id.pop("s1", None) def test_upstream_cache_used_when_no_yaml(self): """Upstream cached instructions are used when no YAML override is set.""" @@ -4706,16 +4707,12 @@ class TestMergeGatewayInitializeInstructions: global_mcp_server_manager, ) - global_mcp_server_manager._upstream_initialize_instructions_by_server_id[ - "s1" - ] = "from upstream" + global_mcp_server_manager._upstream_initialize_instructions_by_server_id["s1"] = "from upstream" try: s = _make_instruction_server(instructions=None) assert self._merge([s]) == "from upstream" finally: - global_mcp_server_manager._upstream_initialize_instructions_by_server_id.pop( - "s1", None - ) + global_mcp_server_manager._upstream_initialize_instructions_by_server_id.pop("s1", None) def test_spec_path_servers_skipped(self): """OpenAPI (spec_path) servers do not contribute instructions.""" @@ -4729,12 +4726,8 @@ class TestMergeGatewayInitializeInstructions: def test_multiple_servers_merged_with_labels(self): """Multiple servers get label-prefixed and separator-joined.""" - s1 = _make_instruction_server( - server_id="a", name="a", alias="Alpha", instructions="instr A" - ) - s2 = _make_instruction_server( - server_id="b", name="b", alias="Beta", instructions="instr B" - ) + s1 = _make_instruction_server(server_id="a", name="a", alias="Alpha", instructions="instr A") + s2 = _make_instruction_server(server_id="b", name="b", alias="Beta", instructions="instr B") result = self._merge([s1, s2]) assert result is not None assert "[Alpha]" in result and "[Beta]" in result @@ -4754,25 +4747,17 @@ class TestMergeGatewayInitializeInstructions: global_mcp_server_manager, ) - global_mcp_server_manager._upstream_initialize_instructions_by_server_id[ - "c" - ] = "cached C" + global_mcp_server_manager._upstream_initialize_instructions_by_server_id["c"] = "cached C" try: - s_yaml = _make_instruction_server( - server_id="a", name="a", alias="A", instructions="yaml A" - ) - s_spec = _make_instruction_server( - server_id="b", name="b", alias="B", spec_path="/spec.json", url=None - ) + s_yaml = _make_instruction_server(server_id="a", name="a", alias="A", instructions="yaml A") + s_spec = _make_instruction_server(server_id="b", name="b", alias="B", spec_path="/spec.json", url=None) s_cached = _make_instruction_server(server_id="c", name="c", alias="C") result = self._merge([s_yaml, s_spec, s_cached]) assert "yaml A" in result assert "cached C" in result assert "[B]" not in result finally: - global_mcp_server_manager._upstream_initialize_instructions_by_server_id.pop( - "c", None - ) + global_mcp_server_manager._upstream_initialize_instructions_by_server_id.pop("c", None) class TestEnsureUpstreamInitializeInstructionsCached: @@ -4784,15 +4769,9 @@ class TestEnsureUpstreamInitializeInstructionsCached: global_mcp_server_manager, ) - server = _make_instruction_server( - server_id="yaml-only", instructions="from yaml" - ) - with patch.object( - global_mcp_server_manager, "_create_mcp_client", AsyncMock() - ) as mock_create: - await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached( - server - ) + server = _make_instruction_server(server_id="yaml-only", instructions="from yaml") + with patch.object(global_mcp_server_manager, "_create_mcp_client", AsyncMock()) as mock_create: + await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached(server) mock_create.assert_not_awaited() @pytest.mark.asyncio @@ -4804,21 +4783,13 @@ class TestEnsureUpstreamInitializeInstructionsCached: ) server = _make_instruction_server(server_id="cached-only", instructions=None) - global_mcp_server_manager._upstream_initialize_instructions_by_server_id[ - "cached-only" - ] = "warm" + global_mcp_server_manager._upstream_initialize_instructions_by_server_id["cached-only"] = "warm" try: - with patch.object( - global_mcp_server_manager, "_create_mcp_client", AsyncMock() - ) as mock_create: - await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached( - server - ) + with patch.object(global_mcp_server_manager, "_create_mcp_client", AsyncMock()) as mock_create: + await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached(server) mock_create.assert_not_awaited() finally: - global_mcp_server_manager._upstream_initialize_instructions_by_server_id.pop( - "cached-only", None - ) + global_mcp_server_manager._upstream_initialize_instructions_by_server_id.pop("cached-only", None) @pytest.mark.asyncio async def test_skips_when_spec_path_set(self): @@ -4828,15 +4799,9 @@ class TestEnsureUpstreamInitializeInstructionsCached: global_mcp_server_manager, ) - server = _make_instruction_server( - server_id="openapi-spec", spec_path="/openapi.json", url=None - ) - with patch.object( - global_mcp_server_manager, "_create_mcp_client", AsyncMock() - ) as mock_create: - await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached( - server - ) + server = _make_instruction_server(server_id="openapi-spec", spec_path="/openapi.json", url=None) + with patch.object(global_mcp_server_manager, "_create_mcp_client", AsyncMock()) as mock_create: + await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached(server) mock_create.assert_not_awaited() @pytest.mark.asyncio @@ -4858,22 +4823,14 @@ class TestEnsureUpstreamInitializeInstructionsCached: AsyncMock(return_value=fake_client), ): try: - await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached( - server - ) + await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached(server) assert ( - global_mcp_server_manager._upstream_initialize_instructions_by_server_id[ - "cold-server" - ] + global_mcp_server_manager._upstream_initialize_instructions_by_server_id["cold-server"] == "upstream says hi" ) finally: - global_mcp_server_manager._upstream_initialize_instructions_by_server_id.pop( - "cold-server", None - ) - global_mcp_server_manager._upstream_initialize_instructions_probed_at.pop( - "cold-server", None - ) + global_mcp_server_manager._upstream_initialize_instructions_by_server_id.pop("cold-server", None) + global_mcp_server_manager._upstream_initialize_instructions_probed_at.pop("cold-server", None) @pytest.mark.asyncio async def test_cooldown_after_empty_upstream_response(self): @@ -4892,27 +4849,13 @@ class TestEnsureUpstreamInitializeInstructionsCached: create = AsyncMock(return_value=fake_client) with patch.object(global_mcp_server_manager, "_create_mcp_client", create): try: - await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached( - server - ) - await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached( - server - ) - assert ( - create.await_count == 1 - ), "Second probe within cooldown must not reconnect to upstream" - assert ( - "empty-server" - not in global_mcp_server_manager._upstream_initialize_instructions_by_server_id - ) - assert ( - "empty-server" - in global_mcp_server_manager._upstream_initialize_instructions_probed_at - ) + await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached(server) + await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached(server) + assert create.await_count == 1, "Second probe within cooldown must not reconnect to upstream" + assert "empty-server" not in global_mcp_server_manager._upstream_initialize_instructions_by_server_id + assert "empty-server" in global_mcp_server_manager._upstream_initialize_instructions_probed_at finally: - global_mcp_server_manager._upstream_initialize_instructions_probed_at.pop( - "empty-server", None - ) + global_mcp_server_manager._upstream_initialize_instructions_probed_at.pop("empty-server", None) @pytest.mark.asyncio async def test_cooldown_after_upstream_failure(self): @@ -4925,35 +4868,19 @@ class TestEnsureUpstreamInitializeInstructionsCached: server = _make_instruction_server(server_id="boom-server", instructions=None) fake_client = MagicMock() - fake_client.run_with_session = AsyncMock( - side_effect=RuntimeError("upstream down") - ) + fake_client.run_with_session = AsyncMock(side_effect=RuntimeError("upstream down")) fake_client._last_initialize_instructions = None create = AsyncMock(return_value=fake_client) with patch.object(global_mcp_server_manager, "_create_mcp_client", create): try: - await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached( - server - ) - await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached( - server - ) - assert ( - create.await_count == 1 - ), "Second probe within cooldown must not reconnect after failure" - assert ( - "boom-server" - not in global_mcp_server_manager._upstream_initialize_instructions_by_server_id - ) - assert ( - "boom-server" - in global_mcp_server_manager._upstream_initialize_instructions_probed_at - ) + await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached(server) + await global_mcp_server_manager._ensure_upstream_initialize_instructions_cached(server) + assert create.await_count == 1, "Second probe within cooldown must not reconnect after failure" + assert "boom-server" not in global_mcp_server_manager._upstream_initialize_instructions_by_server_id + assert "boom-server" in global_mcp_server_manager._upstream_initialize_instructions_probed_at finally: - global_mcp_server_manager._upstream_initialize_instructions_probed_at.pop( - "boom-server", None - ) + global_mcp_server_manager._upstream_initialize_instructions_probed_at.pop("boom-server", None) @pytest.mark.asyncio async def test_reload_resets_probe_cooldown(self): @@ -4962,19 +4889,12 @@ class TestEnsureUpstreamInitializeInstructionsCached: global_mcp_server_manager, ) - global_mcp_server_manager._upstream_initialize_instructions_probed_at[ - "reload-target" - ] = 1.0 + global_mcp_server_manager._upstream_initialize_instructions_probed_at["reload-target"] = 1.0 try: await global_mcp_server_manager.load_servers_from_config({}) - assert ( - "reload-target" - not in global_mcp_server_manager._upstream_initialize_instructions_probed_at - ) + assert "reload-target" not in global_mcp_server_manager._upstream_initialize_instructions_probed_at finally: - global_mcp_server_manager._upstream_initialize_instructions_probed_at.pop( - "reload-target", None - ) + global_mcp_server_manager._upstream_initialize_instructions_probed_at.pop("reload-target", None) class TestGatewayCreateInitializationOptions: @@ -5040,9 +4960,7 @@ class TestGatewayCreateInitializationOptions: ): assert server.create_initialization_options().server_name == "grafana" - assert ( - server.create_initialization_options().server_name == "litellm-mcp-server" - ) + assert server.create_initialization_options().server_name == "litellm-mcp-server" @pytest.mark.asyncio async def test_sse_handler_scopes_server_name_from_single_server_path(self): @@ -5119,9 +5037,7 @@ class TestGatewayCreateInitializationOptions: await handle_sse_mcp(scope, AsyncMock(), AsyncMock()) assert captured["server_name"] == "grafana" - assert ( - server.create_initialization_options().server_name == "litellm-mcp-server" - ) + assert server.create_initialization_options().server_name == "litellm-mcp-server" def test_contextvar_set_injects_instructions(self): """When ContextVar has a value, it appears in InitializationOptions.""" @@ -5239,12 +5155,8 @@ async def test_list_tools_with_legacy_db_m2m_server_resolves_oauth2_flow(): ): mock_manager.get_allowed_mcp_servers = AsyncMock(return_value=["legacy-m2m-id"]) mock_manager.get_mcp_server_by_id = MagicMock(return_value=legacy_server) - mock_manager.filter_server_ids_by_ip_with_info = MagicMock( - return_value=(["legacy-m2m-id"], 0) - ) - mock_manager._get_tools_from_server = AsyncMock( - side_effect=capture_extra_headers - ) + mock_manager.filter_server_ids_by_ip_with_info = MagicMock(return_value=(["legacy-m2m-id"], 0)) + mock_manager._get_tools_from_server = AsyncMock(side_effect=capture_extra_headers) tools = await _get_tools_from_mcp_servers( user_api_key_auth=user_auth, @@ -5336,10 +5248,8 @@ async def test_call_tool_empty_extra_headers_returns_none(): pass # We only care about the captured headers # With P2 fix: extra_headers should be None (not {}) when all headers filtered - assert ( - captured_extra_headers is None - ), "P2 API consistency issue: expected None for empty extra_headers, got: " + str( - captured_extra_headers + assert captured_extra_headers is None, ( + "P2 API consistency issue: expected None for empty extra_headers, got: " + str(captured_extra_headers) ) @@ -5364,9 +5274,7 @@ async def test_probe_upstream_auth_returns_upstream_status(): "litellm.proxy._experimental.mcp_server.server.get_async_httpx_client", return_value=mock_client, ): - status, www_auth = await _probe_upstream_auth( - "http://upstream/mcp", "Bearer some-token" - ) + status, www_auth = await _probe_upstream_auth("http://upstream/mcp", "Bearer some-token") assert status == 401 assert www_auth == 'Bearer realm="test"' @@ -5393,9 +5301,7 @@ async def test_probe_upstream_auth_surfaces_httpx_status_error(): mock_response.status_code = 401 mock_response.headers = {"www-authenticate": 'Bearer realm="test"'} request = httpx.Request("POST", "http://upstream/mcp") - error = httpx.HTTPStatusError( - message="401 Unauthorized", request=request, response=mock_response - ) + error = httpx.HTTPStatusError(message="401 Unauthorized", request=request, response=mock_response) mock_client = MagicMock() mock_client.post = AsyncMock(side_effect=error) @@ -5404,9 +5310,7 @@ async def test_probe_upstream_auth_surfaces_httpx_status_error(): "litellm.proxy._experimental.mcp_server.server.get_async_httpx_client", return_value=mock_client, ): - status, www_auth = await _probe_upstream_auth( - "http://upstream/mcp", "Bearer some-token" - ) + status, www_auth = await _probe_upstream_auth("http://upstream/mcp", "Bearer some-token") assert status == 401 assert www_auth == 'Bearer realm="test"' @@ -5424,9 +5328,7 @@ async def test_probe_upstream_auth_fails_open_on_network_error(): "litellm.proxy._experimental.mcp_server.server.get_async_httpx_client", return_value=mock_client, ): - status, www_auth = await _probe_upstream_auth( - "http://upstream/mcp", "Bearer some-token" - ) + status, www_auth = await _probe_upstream_auth("http://upstream/mcp", "Bearer some-token") assert status == 200 assert www_auth is None @@ -6349,9 +6251,7 @@ class TestProxyExceptionToHttpException: from litellm.proxy._types import ProxyException http_exc = _proxy_exception_to_http_exception( - ProxyException( - message="Forbidden", type="auth_error", param="key", code=403 - ) + ProxyException(message="Forbidden", type="auth_error", param="key", code=403) ) assert http_exc.status_code == 403 @@ -6415,10 +6315,7 @@ class TestStreamableHttpAuthErrorMapping: assert exc_info.value.status_code == 401 assert exc_info.value.headers["WWW-Authenticate"] == "Bearer" # Must not have emitted a 500 body via the generic catch-all. - assert not any( - m.get("type") == "http.response.start" and m.get("status") == 500 - for m in sent - ) + assert not any(m.get("type") == "http.response.start" and m.get("status") == 500 for m in sent) @pytest.mark.asyncio async def test_sse_propagates_proxy_exception_as_401(self): @@ -6458,10 +6355,7 @@ class TestStreamableHttpAuthErrorMapping: assert exc_info.value.status_code == 401 assert exc_info.value.headers["WWW-Authenticate"] == "Bearer" - assert not any( - m.get("type") == "http.response.start" and m.get("status") == 500 - for m in sent - ) + assert not any(m.get("type") == "http.response.start" and m.get("status") == 500 for m in sent) class TestMCPMetaTraceCarrier: @@ -6598,6 +6492,308 @@ async def test_get_active_submitted_mcp_server_ids_for_user_empty_user_id_skips_ prisma_client.db.litellm_mcpservertable.find_many.assert_not_awaited() +# --------------------------------------------------------------------------- # +# MCP tool-call isError failure logging +# --------------------------------------------------------------------------- # + + +def _call_tool_result(is_error: bool, text: str) -> CallToolResult: + return CallToolResult(content=[TextContent(type="text", text=text)], isError=is_error) + + +def _mock_mcp_logging_obj() -> MagicMock: + logging_obj = MagicMock() + logging_obj.model_call_details = {} + logging_obj.async_post_mcp_tool_call_hook = AsyncMock() + logging_obj.async_success_handler = AsyncMock() + logging_obj.async_failure_handler = AsyncMock() + return logging_obj + + +def test_extract_mcp_tool_result_error_message(): + from litellm.proxy._experimental.mcp_server.utils import ( + extract_mcp_tool_result_error_message, + ) + + assert extract_mcp_tool_result_error_message(_call_tool_result(True, "boom")) == "boom" + assert extract_mcp_tool_result_error_message(_call_tool_result(False, "ok")) is None + assert ( + extract_mcp_tool_result_error_message(CallToolResult(content=[], isError=True)) + == "MCP tool call returned isError=true" + ) + assert ( + extract_mcp_tool_result_error_message({"isError": True, "content": [{"type": "text", "text": "denied"}]}) + == "denied" + ) + assert extract_mcp_tool_result_error_message({"isError": False, "content": []}) is None + assert extract_mcp_tool_result_error_message({}) is None + + +@pytest.mark.asyncio +async def test_fire_mcp_tool_call_logging_iserror_logs_failure(): + """Regression test: a CallToolResult with isError=True must go + down the failure logging path (async_failure_handler + post_call_failure_hook), + never async_success_handler.""" + from litellm.proxy._experimental.mcp_server.server import ( + _fire_mcp_tool_call_logging, + ) + from litellm.proxy._experimental.mcp_server.exceptions import MCPToolResultError + + logging_obj = _mock_mcp_logging_obj() + proxy_logging_mock = MagicMock() + proxy_logging_mock.post_call_failure_hook = AsyncMock() + user_auth = UserAPIKeyAuth(api_key="test-key", user_id="test-user") + + with patch("litellm.proxy.proxy_server.proxy_logging_obj", proxy_logging_mock): + await _fire_mcp_tool_call_logging( + logging_obj=logging_obj, + result=_call_tool_result(True, "upstream exploded"), + start_time=datetime.now(), + end_time=datetime.now(), + user_api_key_auth=user_auth, + request_data={"litellm_call_id": "cid"}, + ) + + logging_obj.async_success_handler.assert_not_awaited() + logging_obj.failure_handler.assert_called_once() + logging_obj.async_failure_handler.assert_awaited_once() + tool_error = logging_obj.async_failure_handler.await_args.args[0] + assert isinstance(tool_error, MCPToolResultError) + assert str(tool_error) == "upstream exploded" + logging_obj.has_run_logging.assert_any_call(event_type="sync_success") + logging_obj.has_run_logging.assert_any_call(event_type="async_success") + proxy_logging_mock.post_call_failure_hook.assert_awaited_once() + hook_kwargs = proxy_logging_mock.post_call_failure_hook.await_args.kwargs + assert hook_kwargs["route"] == "/mcp/call_tool" + assert hook_kwargs["original_exception"] is tool_error + assert hook_kwargs["user_api_key_dict"] is user_auth + logging_obj.async_post_mcp_tool_call_hook.assert_awaited_once() + + +@pytest.mark.asyncio +async def test_fire_mcp_tool_call_logging_success_path_unchanged(): + """isError=False must keep today's behavior: success handler fires, no + failure logging, no post_call_failure_hook.""" + from litellm.proxy._experimental.mcp_server.server import ( + _fire_mcp_tool_call_logging, + ) + + logging_obj = _mock_mcp_logging_obj() + proxy_logging_mock = MagicMock() + proxy_logging_mock.post_call_failure_hook = AsyncMock() + result = _call_tool_result(False, "all good") + + with patch("litellm.proxy.proxy_server.proxy_logging_obj", proxy_logging_mock): + await _fire_mcp_tool_call_logging( + logging_obj=logging_obj, + result=result, + start_time=datetime.now(), + end_time=datetime.now(), + user_api_key_auth=UserAPIKeyAuth(api_key="test-key", user_id="test-user"), + request_data={}, + ) + + logging_obj.async_success_handler.assert_awaited_once() + assert logging_obj.async_success_handler.await_args.kwargs["result"] is result + logging_obj.async_failure_handler.assert_not_awaited() + logging_obj.failure_handler.assert_not_called() + proxy_logging_mock.post_call_failure_hook.assert_not_awaited() + + +@pytest.mark.asyncio +async def test_fire_mcp_tool_call_logging_iserror_without_auth_skips_failure_hook(): + """Without a UserAPIKeyAuth the failure handlers still fire but the proxy + post_call_failure_hook (which requires one) is skipped.""" + from litellm.proxy._experimental.mcp_server.server import ( + _fire_mcp_tool_call_logging, + ) + + logging_obj = _mock_mcp_logging_obj() + proxy_logging_mock = MagicMock() + proxy_logging_mock.post_call_failure_hook = AsyncMock() + + with patch("litellm.proxy.proxy_server.proxy_logging_obj", proxy_logging_mock): + await _fire_mcp_tool_call_logging( + logging_obj=logging_obj, + result={"isError": True, "content": [{"type": "text", "text": "denied"}]}, + start_time=datetime.now(), + end_time=datetime.now(), + ) + + logging_obj.async_success_handler.assert_not_awaited() + logging_obj.async_failure_handler.assert_awaited_once() + assert str(logging_obj.async_failure_handler.await_args.args[0]) == "denied" + proxy_logging_mock.post_call_failure_hook.assert_not_awaited() + + +@pytest.mark.asyncio +async def test_fire_mcp_tool_call_logging_strips_credentials_from_failure_hook(): + """Credential-bearing request_data fields (raw request headers, upstream MCP + auth headers, OAuth tokens) must never reach post_call_failure_hook + callbacks; non-credential fields must survive untouched.""" + from litellm.proxy._experimental.mcp_server.server import ( + _fire_mcp_tool_call_logging, + ) + + logging_obj = _mock_mcp_logging_obj() + proxy_logging_mock = MagicMock() + proxy_logging_mock.post_call_failure_hook = AsyncMock() + user_auth = UserAPIKeyAuth(api_key="test-key", user_id="test-user") + request_data = { + "name": "explode", + "litellm_call_id": "cid", + "raw_headers": {"authorization": "Bearer sk-caller-secret"}, + "mcp_auth_header": "upstream-secret", + "mcp_server_auth_headers": {"srv": {"authorization": "Bearer srv-secret"}}, + "oauth2_headers": {"authorization": "Bearer oauth-secret"}, + "user_api_key_auth": user_auth, + } + + with patch("litellm.proxy.proxy_server.proxy_logging_obj", proxy_logging_mock): + await _fire_mcp_tool_call_logging( + logging_obj=logging_obj, + result=_call_tool_result(True, "boom"), + start_time=datetime.now(), + end_time=datetime.now(), + user_api_key_auth=user_auth, + request_data=request_data, + ) + + proxy_logging_mock.post_call_failure_hook.assert_awaited_once() + hook_request_data = proxy_logging_mock.post_call_failure_hook.await_args.kwargs["request_data"] + assert hook_request_data == {"name": "explode", "litellm_call_id": "cid"} + assert "secret" not in str(hook_request_data) + + +def _real_mcp_logging_obj(call_id: str): + from litellm.litellm_core_utils.litellm_logging import Logging + + start_time = datetime.now() + logging_obj = Logging( + model="MCP: weather/get_forecast", + messages=[{"role": "user", "content": "tool call"}], + stream=False, + call_type="call_mcp_tool", + start_time=start_time, + litellm_call_id=call_id, + function_id="test-fn", + ) + logging_obj.update_environment_variables( + model="MCP: weather/get_forecast", + user="", + optional_params={}, + litellm_params={"api_base": ""}, + ) + logging_obj.model_call_details["mcp_tool_call_metadata"] = { + "name": "get_forecast", + "arguments": {"city": "Paris"}, + "mcp_server_name": "weather", + } + return logging_obj, start_time + + +@pytest.mark.asyncio +async def test_fire_mcp_tool_call_logging_iserror_builds_failure_payload(monkeypatch): + """The standard logging payload for an isError=True result must carry + status='failure' with the tool's error text, so OTel (whose _parse_error + keys off status) marks the MCP span ERROR.""" + import litellm + from litellm.proxy._experimental.mcp_server.server import ( + _fire_mcp_tool_call_logging, + ) + + monkeypatch.setattr(litellm, "failure_callback", []) + monkeypatch.setattr(litellm, "_async_failure_callback", []) + monkeypatch.setattr(litellm, "success_callback", []) + monkeypatch.setattr(litellm, "_async_success_callback", []) + + logging_obj, start_time = _real_mcp_logging_obj("test-mcp-iserror-payload") + + await _fire_mcp_tool_call_logging( + logging_obj=logging_obj, + result=_call_tool_result(True, "upstream exploded"), + start_time=start_time, + end_time=datetime.now(), + ) + + payload = logging_obj.model_call_details["standard_logging_object"] + assert payload["status"] == "failure" + assert payload["error_str"] == "upstream exploded" + assert payload["error_information"]["error_class"] == "MCPToolResultError" + assert payload["metadata"]["mcp_tool_call_metadata"]["name"] == "get_forecast" + + +@pytest.mark.asyncio +async def test_fire_mcp_tool_call_logging_success_builds_success_payload(monkeypatch): + """isError=False still produces a status='success' payload.""" + import litellm + from litellm.proxy._experimental.mcp_server.server import ( + _fire_mcp_tool_call_logging, + ) + + monkeypatch.setattr(litellm, "failure_callback", []) + monkeypatch.setattr(litellm, "_async_failure_callback", []) + monkeypatch.setattr(litellm, "success_callback", []) + monkeypatch.setattr(litellm, "_async_success_callback", []) + + logging_obj, start_time = _real_mcp_logging_obj("test-mcp-success-payload") + + await _fire_mcp_tool_call_logging( + logging_obj=logging_obj, + result=_call_tool_result(False, "all good"), + start_time=start_time, + end_time=datetime.now(), + ) + + payload = logging_obj.model_call_details["standard_logging_object"] + assert payload["status"] == "success" + + +@pytest.mark.asyncio +async def test_fire_mcp_tool_call_logging_iserror_emits_otel_error_span(monkeypatch): + """End-to-end regression for the OTel symptom: an isError=True tool + result must reach OTel as an MCP span with StatusCode.ERROR and the tool's + error message, while isError=False stays non-error.""" + pytest.importorskip("opentelemetry") + from opentelemetry.sdk.trace.export.in_memory_span_exporter import ( + InMemorySpanExporter, + ) + from opentelemetry.trace.status import StatusCode + + import litellm + from litellm.integrations.otel import OpenTelemetryV2Config + from litellm.integrations.otel.logger import OpenTelemetryV2 + from litellm.integrations.otel.plumbing import providers + from litellm.proxy._experimental.mcp_server.server import ( + _fire_mcp_tool_call_logging, + ) + + cfg = OpenTelemetryV2Config(exporter="in_memory", legacy_compat=False) + exporter = InMemorySpanExporter() + tracer_provider = providers.build_tracer_provider(cfg, exporter=exporter) + otel_logger = OpenTelemetryV2(config=cfg, tracer_provider=tracer_provider) + + monkeypatch.setattr(litellm, "failure_callback", []) + monkeypatch.setattr(litellm, "_async_failure_callback", [otel_logger]) + monkeypatch.setattr(litellm, "success_callback", []) + monkeypatch.setattr(litellm, "_async_success_callback", [otel_logger]) + + logging_obj, start_time = _real_mcp_logging_obj("test-mcp-iserror-otel") + + await _fire_mcp_tool_call_logging( + logging_obj=logging_obj, + result=_call_tool_result(True, "upstream exploded"), + start_time=start_time, + end_time=datetime.now(), + ) + + (span,) = exporter.get_finished_spans() + assert span.name == "tools/call get_forecast" + assert span.status.status_code is StatusCode.ERROR + assert span.attributes["error.type"] == "MCPToolResultError" + assert "upstream exploded" in (span.status.description or "") + + @pytest.mark.asyncio async def test_call_tool_with_legacy_db_m2m_server_resolves_oauth2_flow(): """ @@ -6658,3 +6854,28 @@ async def test_call_tool_with_legacy_db_m2m_server_resolves_oauth2_flow(): resolved = captured_servers["allowed"] assert resolved and resolved[0].oauth2_flow == "client_credentials" assert resolved[0].has_client_credentials is True + + +@pytest.mark.parametrize( + "url, expected", + [ + # only the origin may be logged: userinfo, query, fragment, and the PATH are all stripped, + # because hosted MCP servers routinely embed the credential in the path (e.g. /mcp/s/) + ("https://user:s3cr3t@mcp.example.com/mcp?token=abcd1234&x=1", "https://mcp.example.com"), + ("https://mcp.example.com/mcp#frag", "https://mcp.example.com"), + ("https://host:8443/a/b?q=1", "https://host:8443"), + ("https://mcp.zapier.com/api/mcp/s/NDgzcret-token/mcp", "https://mcp.zapier.com"), + ("https://mcp.notion.com/mcp", "https://mcp.notion.com"), + (None, None), + ("", None), + ("not a url", None), + ("http://[::1", None), + ], +) +def test_redact_mcp_resource_url_strips_credentials(url, expected): + """The MCP tool-call log records the upstream resource, so the URL must be redacted to + scheme+host+path: userinfo, query string, and fragment (which can carry embedded tokens or + secret parameters) must never reach spend-log metadata or logging callbacks.""" + from litellm.proxy._experimental.mcp_server.server import _redact_mcp_resource_url + + assert _redact_mcp_resource_url(url) == expected diff --git a/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_server_manager.py b/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_server_manager.py index a423f7c8b83..a18e1ac2c44 100644 --- a/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_server_manager.py +++ b/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_server_manager.py @@ -833,6 +833,39 @@ class TestMCPServerManager: assert spec is not None and isinstance(spec.config, TokenExchangeConfig) assert spec.config.profile == "entra_obo" + @pytest.mark.asyncio + @pytest.mark.parametrize("auth_type", [MCPAuth.true_passthrough, MCPAuth.oauth_delegate]) + async def test_build_from_table_discovers_upstream_oauth_for_client_forwarded_modes(self, auth_type): + """The gateway's relayed authorize flow (used by the browser-only Authorize) needs the + upstream's authorization_url on the registry entry, and these rows never persist one, so + the DB build must discover it the same way oauth2 rows do.""" + from types import SimpleNamespace + + manager = MCPServerManager() + row = LiteLLM_MCPServerTable( + server_id="cf-db-1", + alias="cf_db", + description="client-forwarded from db", + url="https://up.example.com/mcp", + transport=MCPTransport.http, + auth_type=auth_type, + created_at=datetime.now(), + updated_at=datetime.now(), + ) + + metadata = SimpleNamespace( + authorization_url="https://idp.example.com/authorize", + token_url="https://idp.example.com/token", + registration_url="https://idp.example.com/register", + scopes=None, + ) + with patch.object(manager, "_descovery_metadata", new=AsyncMock(return_value=metadata)) as mock_discovery: + built = await manager.build_mcp_server_from_table(row, credentials_are_encrypted=False) + + mock_discovery.assert_awaited_once() + assert built.authorization_url == "https://idp.example.com/authorize" + assert built.token_url == "https://idp.example.com/token" + async def _capture_subject_token(self, call) -> Optional[str]: """Run a manager method (via ``call(manager)``) and return the subject_token it threaded into ``_create_mcp_client``.""" @@ -1267,6 +1300,341 @@ class TestMCPServerManager: assert captured_extra_headers == {"Authorization": "Bearer upstream-oauth-bearer"} + async def _capture_call_extra_headers(self, server, oauth2_headers, raw_headers, user_api_key_auth): + manager = MCPServerManager() + mock_client = AsyncMock() + mock_client.call_tool = AsyncMock(return_value=CallToolResult(content=[], isError=False)) + captured = {"extra_headers": "unset"} + + async def capture_create_mcp_client( + server, mcp_auth_header, extra_headers, stdio_env, subject_token=None, **kwargs + ): # pragma: no cover - helper + captured["extra_headers"] = extra_headers + return mock_client + + manager._create_mcp_client = AsyncMock(side_effect=capture_create_mcp_client) + await manager._call_regular_mcp_tool( + mcp_server=server, + original_tool_name="tool", + arguments={}, + tasks=[], + mcp_auth_header=None, + mcp_server_auth_headers=None, + oauth2_headers=oauth2_headers, + raw_headers=raw_headers, + proxy_logging_obj=None, + user_api_key_auth=user_api_key_auth, + ) + return captured["extra_headers"] + + @pytest.mark.asyncio + async def test_call_regular_mcp_tool_true_passthrough_forwards_authorization(self): + from litellm.proxy._types import UserAPIKeyAuth + + server = MCPServer( + server_id="server-true-passthrough", + name="tp-server", + url="https://example.com", + transport=MCPTransport.http, + auth_type=MCPAuth.true_passthrough, + ) + extra_headers = await self._capture_call_extra_headers( + server, + oauth2_headers={"Authorization": "Bearer upstream-token"}, + raw_headers={"authorization": "Bearer upstream-token"}, + user_api_key_auth=UserAPIKeyAuth(api_key=None), + ) + assert extra_headers == {"Authorization": "Bearer upstream-token"} + + @pytest.mark.asyncio + async def test_call_regular_mcp_tool_oauth_delegate_forwards_separate_authorization(self): + from litellm.proxy._types import UserAPIKeyAuth + + server = MCPServer( + server_id="server-oauth-delegate", + name="od-server", + url="https://example.com", + transport=MCPTransport.http, + auth_type=MCPAuth.oauth_delegate, + ) + extra_headers = await self._capture_call_extra_headers( + server, + oauth2_headers={"Authorization": "Bearer upstream-token"}, + raw_headers={ + "x-litellm-api-key": "Bearer sk-litellm-key", + "authorization": "Bearer upstream-token", + }, + user_api_key_auth=UserAPIKeyAuth(api_key="sk-litellm-key"), + ) + assert extra_headers == {"Authorization": "Bearer upstream-token"} + + @pytest.mark.asyncio + async def test_call_regular_mcp_tool_oauth_delegate_never_forwards_admission_key(self): + from litellm.proxy._types import UserAPIKeyAuth + + server = MCPServer( + server_id="server-oauth-delegate-leak", + name="od-server", + url="https://example.com", + transport=MCPTransport.http, + auth_type=MCPAuth.oauth_delegate, + ) + extra_headers = await self._capture_call_extra_headers( + server, + oauth2_headers={"Authorization": "Bearer sk-litellm-key"}, + raw_headers={"authorization": "Bearer sk-litellm-key"}, + user_api_key_auth=UserAPIKeyAuth(api_key="sk-litellm-key"), + ) + assert not extra_headers or "authorization" not in {k.lower() for k in extra_headers} + + def test_should_strip_caller_authorization_new_modes(self): + from litellm.proxy._types import UserAPIKeyAuth + + true_passthrough = MCPServer( + server_id="tp", + name="tp", + url="https://example.com", + transport=MCPTransport.http, + auth_type=MCPAuth.true_passthrough, + ) + assert ( + _should_strip_caller_authorization( + mcp_server=true_passthrough, + raw_headers={"authorization": "Bearer upstream"}, + user_api_key_auth=UserAPIKeyAuth(api_key=None), + ) + is False + ) + + oauth_delegate = MCPServer( + server_id="od", + name="od", + url="https://example.com", + transport=MCPTransport.http, + auth_type=MCPAuth.oauth_delegate, + ) + assert ( + _should_strip_caller_authorization( + mcp_server=oauth_delegate, + raw_headers={ + "x-litellm-api-key": "Bearer sk-litellm-key", + "authorization": "Bearer upstream", + }, + user_api_key_auth=UserAPIKeyAuth(api_key="sk-litellm-key"), + ) + is False + ) + assert ( + _should_strip_caller_authorization( + mcp_server=oauth_delegate, + raw_headers={"authorization": "Bearer sk-litellm-key"}, + user_api_key_auth=UserAPIKeyAuth(api_key="sk-litellm-key"), + ) + is True + ) + + def test_should_strip_authorization_for_oauth_delegate_admitted_via_jwt_without_api_key(self): + """JWT / SSO / OIDC / session admission yields a UserAPIKeyAuth with a user_id but + api_key=None; the caller's Authorization was that credential and must be stripped for + oauth_delegate when no separate x-litellm-api-key carried admission (LIT-3794-class leak).""" + from litellm.proxy._types import UserAPIKeyAuth + + oauth_delegate = MCPServer( + server_id="od-jwt", + name="od", + url="https://example.com", + transport=MCPTransport.http, + auth_type=MCPAuth.oauth_delegate, + ) + assert ( + _should_strip_caller_authorization( + mcp_server=oauth_delegate, + raw_headers={"authorization": "Bearer eyJ-idp-jwt"}, + user_api_key_auth=UserAPIKeyAuth(user_id="alice", api_key=None), + ) + is True + ) + assert ( + _should_strip_caller_authorization( + mcp_server=oauth_delegate, + raw_headers={ + "x-litellm-api-key": "Bearer sk-1234", + "authorization": "Bearer upstream", + }, + user_api_key_auth=UserAPIKeyAuth(user_id="alice", api_key=None), + ) + is False + ) + + @pytest.mark.asyncio + async def test_call_regular_mcp_tool_oauth_delegate_never_forwards_jwt_admission(self): + from litellm.proxy._types import UserAPIKeyAuth + + server = MCPServer( + server_id="od-jwt-e2e", + name="od", + url="https://example.com", + transport=MCPTransport.http, + auth_type=MCPAuth.oauth_delegate, + ) + extra_headers = await self._capture_call_extra_headers( + server, + oauth2_headers={"Authorization": "Bearer eyJ-idp-jwt"}, + raw_headers={"authorization": "Bearer eyJ-idp-jwt"}, + user_api_key_auth=UserAPIKeyAuth(user_id="alice", api_key=None), + ) + assert not extra_headers or "authorization" not in {k.lower() for k in extra_headers} + + def test_new_passthrough_modes_require_per_user_auth(self): + for auth_type in (MCPAuth.true_passthrough, MCPAuth.oauth_delegate): + server = MCPServer( + server_id="s", + name="s", + url="https://example.com", + transport=MCPTransport.http, + auth_type=auth_type, + ) + assert server.requires_per_user_auth is True + + @pytest.mark.asyncio + async def test_create_mcp_client_forwarded_modes_use_the_passthrough_arm(self): + manager = MCPServerManager() + server = MCPServer( + server_id="tp-egress", + name="tp", + url="https://example.com", + transport=MCPTransport.http, + auth_type=MCPAuth.true_passthrough, + ) + with ( + patch( + "litellm.proxy._experimental.mcp_server.mcp_server_manager.resolve_mcp_auth", + new_callable=AsyncMock, + ) as mock_resolve, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.MCPClient") as mock_client_cls, + ): + await manager._create_mcp_client(server=server, extra_headers={"Authorization": "Bearer upstream-token"}) + mock_resolve.assert_not_awaited() + kwargs = mock_client_cls.call_args.kwargs + emitted = httpx.Request("GET", "https://example.com/mcp") + flow = kwargs["resolved_auth"].auth_flow(emitted) + next(flow) + flow.close() + assert emitted.headers["Authorization"] == "Bearer upstream-token" + assert not kwargs["extra_headers"] or "authorization" not in {k.lower() for k in kwargs["extra_headers"]} + + @staticmethod + def _emitted_authorization(mock_client_cls) -> str: + kwargs = mock_client_cls.call_args.kwargs + emitted = httpx.Request("GET", "https://example.com/mcp") + flow = kwargs["resolved_auth"].auth_flow(emitted) + next(flow) + flow.close() + return emitted.headers["Authorization"] + + @pytest.mark.asyncio + @pytest.mark.parametrize( + "per_server_header", + ["Bearer per-server-token", {"Authorization": "Bearer per-server-token"}], + ) + async def test_create_mcp_client_passthrough_prefers_per_server_token(self, per_server_header): + """A per-server x-mcp-{alias}-authorization value is the explicit one-token-one-server + binding, so it must win over the request-wide Authorization and reach the upstream + verbatim through the passthrough arm.""" + manager = MCPServerManager() + server = MCPServer( + server_id="tp-per-server", + name="tp", + url="https://example.com", + transport=MCPTransport.http, + auth_type=MCPAuth.true_passthrough, + ) + with ( + patch( + "litellm.proxy._experimental.mcp_server.mcp_server_manager.resolve_mcp_auth", + new_callable=AsyncMock, + ) as mock_resolve, + patch("litellm.proxy._experimental.mcp_server.mcp_server_manager.MCPClient") as mock_client_cls, + ): + await manager._create_mcp_client( + server=server, + mcp_auth_header=per_server_header, + extra_headers={"Authorization": "Bearer global-token"}, + ) + mock_resolve.assert_not_awaited() + assert self._emitted_authorization(mock_client_cls) == "Bearer per-server-token" + kwargs = mock_client_cls.call_args.kwargs + assert not kwargs["extra_headers"] or "authorization" not in {k.lower() for k in kwargs["extra_headers"]} + + def test_consumes_caller_authorization_per_mode(self): + from litellm.proxy._experimental.mcp_server.mcp_server_manager import ( + _consumes_caller_authorization, + ) + + def build(**kwargs) -> MCPServer: + return MCPServer( + server_id="s", + name="s", + url="https://example.com", + transport=MCPTransport.http, + **kwargs, + ) + + assert _consumes_caller_authorization(build(auth_type=MCPAuth.true_passthrough)) is True + assert _consumes_caller_authorization(build(auth_type=MCPAuth.oauth_delegate)) is True + assert ( + _consumes_caller_authorization( + build(auth_type=MCPAuth.none, extra_headers=["Authorization"], oauth_passthrough=True) + ) + is True + ) + assert _consumes_caller_authorization(build(auth_type=MCPAuth.oauth2, delegate_auth_to_upstream=True)) is True + assert _consumes_caller_authorization(build(auth_type=MCPAuth.api_key, authentication_token="x")) is False + assert ( + _consumes_caller_authorization( + build( + auth_type=MCPAuth.oauth2, + delegate_auth_to_upstream=True, + oauth2_flow="client_credentials", + token_url="https://idp/token", + ) + ) + is False + ) + + def test_caller_authorization_fans_out_only_with_second_consumer(self): + from litellm.proxy._experimental.mcp_server.mcp_server_manager import ( + _caller_authorization_fans_out, + ) + + delegate = MCPServer( + server_id="od", + name="od", + url="https://example.com", + transport=MCPTransport.http, + auth_type=MCPAuth.oauth_delegate, + ) + second = MCPServer( + server_id="tp", + name="tp", + url="https://example.com", + transport=MCPTransport.http, + auth_type=MCPAuth.true_passthrough, + ) + static_server = MCPServer( + server_id="static", + name="static", + url="https://example.com", + transport=MCPTransport.http, + auth_type=MCPAuth.api_key, + authentication_token="x", + ) + + assert _caller_authorization_fans_out(delegate, None) is False + assert _caller_authorization_fans_out(delegate, [delegate]) is False + assert _caller_authorization_fans_out(delegate, [delegate, static_server]) is False + assert _caller_authorization_fans_out(delegate, [delegate, second]) is True + @pytest.mark.asyncio async def test_get_prompts_from_server_success(self): """Ensure prompts are fetched and prefixed when requested.""" @@ -2959,9 +3327,35 @@ class TestMCPServerManager: await manager.invalidate_user_oauth_token_cache("alice", "srv-1") assert store.invalidations == [("alice", "srv-1")] + @pytest.mark.asyncio + async def test_invalidate_user_oauth_token_cache_drops_legacy_cache_too(self): + """A per-user token can be served from the legacy per-user token cache as well as the v2 + store; the shared invalidation must evict both, or the path not evicted keeps serving a + token minted for a replaced credential row until its TTL.""" + + class _Store: + async def fetch(self, user_id: str, server_id: str): + return None + + async def invalidate(self, user_id: str, server_id: str) -> None: + return None + + class _LegacyCache: + def __init__(self) -> None: + self.deletes: list[tuple[str, str]] = [] + + async def delete(self, user_id: str, server_id: str) -> None: + self.deletes.append((user_id, server_id)) + + legacy_cache = _LegacyCache() + manager = MCPServerManager(per_user_oauth_token_store=_Store(), per_user_token_cache=legacy_cache) + await manager.invalidate_user_oauth_token_cache("alice", "srv-1") + assert legacy_cache.deletes == [("alice", "srv-1")] + @pytest.mark.asyncio async def test_invalidate_user_oauth_token_cache_swallows_store_errors(self): - """A cache-drop failure must not fail the credential write that triggered it.""" + """A cache-drop failure must not fail the credential write that triggered it, and the + legacy cache must still be evicted after the v2 store drop fails.""" class _Store: async def fetch(self, user_id: str, server_id: str): @@ -2970,7 +3364,35 @@ class TestMCPServerManager: async def invalidate(self, user_id: str, server_id: str) -> None: raise RuntimeError("redis down") - manager = MCPServerManager(per_user_oauth_token_store=_Store()) + class _LegacyCache: + def __init__(self) -> None: + self.deletes: list[tuple[str, str]] = [] + + async def delete(self, user_id: str, server_id: str) -> None: + self.deletes.append((user_id, server_id)) + + legacy_cache = _LegacyCache() + manager = MCPServerManager(per_user_oauth_token_store=_Store(), per_user_token_cache=legacy_cache) + await manager.invalidate_user_oauth_token_cache("alice", "srv-1") + assert legacy_cache.deletes == [("alice", "srv-1")] + + @pytest.mark.asyncio + async def test_invalidate_user_oauth_token_cache_swallows_legacy_cache_errors(self): + """The legacy cache drop is best-effort like the v2 drop: a failure must be logged, never + raised into the credential write that triggered the invalidation.""" + + class _Store: + async def fetch(self, user_id: str, server_id: str): + return None + + async def invalidate(self, user_id: str, server_id: str) -> None: + return None + + class _RaisingLegacyCache: + async def delete(self, user_id: str, server_id: str) -> None: + raise RuntimeError("redis down") + + manager = MCPServerManager(per_user_oauth_token_store=_Store(), per_user_token_cache=_RaisingLegacyCache()) await manager.invalidate_user_oauth_token_cache("alice", "srv-1") @pytest.mark.asyncio diff --git a/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_stale_session.py b/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_stale_session.py index 76614a35b53..da183e8d02a 100644 --- a/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_stale_session.py +++ b/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_stale_session.py @@ -713,6 +713,9 @@ async def test_handle_streamable_http_mcp_delegated_server_surfaces_upstream_cha delegated_server.auth_type = MCPAuth.oauth2 delegated_server.delegate_auth_to_upstream = True delegated_server.needs_user_oauth_token = True + delegated_server.is_oauth_passthrough = False + delegated_server.is_oauth_delegate = False + delegated_server.is_true_passthrough = False delegated_server.server_id = "delegated-oauth-server" upstream_challenge = 'Bearer resource_metadata="https://upstream.example.com/.well-known/oauth-protected-resource"' @@ -1048,3 +1051,424 @@ async def test_handle_streamable_http_mcp_token_exchange_without_subject_returns assert "/.well-known/oauth-protected-resource" in challenge assert challenge.split('resource_metadata="', 1)[1].split('"', 1)[0].endswith("/mcp/obo_server") assert 'error="invalid_token"' in challenge + + +def _passthrough_mode_scope(server_name: str, extra_headers=None): + headers = [ + (b"content-type", b"application/json"), + (b"host", b"litellm.example.com"), + ] + list(extra_headers or []) + return { + "type": "http", + "method": "POST", + "path": f"/mcp/{server_name}", + "_original_path": f"/{server_name}/mcp", + "scheme": "https", + "query_string": b"", + "root_path": "", + "server": ("litellm.example.com", 443), + "headers": headers, + } + + +def _build_passthrough_mode_server(server_name: str, auth_type): + from litellm.proxy._types import MCPTransport + from litellm.types.mcp_server.mcp_server_manager import MCPServer + + return MCPServer( + server_id=f"{server_name}-id", + name=server_name, + server_name=server_name, + alias=server_name, + url="https://upstream.example.com/mcp", + transport=MCPTransport.http, + auth_type=auth_type, + ) + + +@pytest.mark.asyncio +async def test_handle_streamable_http_mcp_oauth_delegate_without_token_returns_gateway_proxied_401(): + """oauth_delegate is admitted with the LiteLLM key but still owns upstream + OAuth. With no forwarded upstream token the gateway must challenge with the + proxied resource_metadata (which advertises the upstream IdP), never the + gateway authorization_uri and never a silent 200.""" + from fastapi import HTTPException + + try: + from litellm.proxy._experimental.mcp_server.server import ( + handle_streamable_http_mcp, + session_manager_stateful, + ) + except ImportError: + pytest.skip("MCP server not available") + + scope = _passthrough_mode_scope("od_server") + receive = AsyncMock( + return_value={ + "type": "http.request", + "body": b'{"jsonrpc":"2.0","id":1,"method":"initialize","params":{}}', + "more_body": False, + } + ) + send = AsyncMock() + user_auth = MagicMock() + user_auth.user_id = "u1" + od_server = _build_passthrough_mode_server("od_server", MCPAuth.oauth_delegate) + + with ( + patch( + "litellm.proxy._experimental.mcp_server.server.extract_mcp_auth_context", + new_callable=AsyncMock, + return_value=(user_auth, None, ["od_server"], None, None, None), + ), + patch("litellm.proxy._experimental.mcp_server.server.set_auth_context"), + patch( + "litellm.proxy._experimental.mcp_server.server._SESSION_MANAGERS_INITIALIZED", + True, + ), + patch( + "litellm.proxy._experimental.mcp_server.server.global_mcp_server_manager.get_mcp_server_by_name", + return_value=od_server, + ), + patch.object( + session_manager_stateful, + "handle_request", + new_callable=AsyncMock, + ) as mock_handle_request, + ): + with pytest.raises(HTTPException) as exc_info: + await handle_streamable_http_mcp(scope, receive, send) + + assert mock_handle_request.await_count == 0 + assert exc_info.value.status_code == 401 + challenge = exc_info.value.headers["www-authenticate"] + assert "resource_metadata=" in challenge + assert "authorization_uri=" not in challenge + assert "/.well-known/oauth-protected-resource/od_server/mcp" in challenge + + +@pytest.mark.asyncio +async def test_handle_streamable_http_mcp_oauth_delegate_with_forwarded_token_skips_challenge(): + """When the oauth_delegate caller carries both the LiteLLM key and a separate + upstream Authorization, the gateway must forward to the session manager, not + re-challenge. Guards the ``_get_forwarded_auth_from_scope(...) is None`` + condition: dropping it would 401 even a fully-authenticated request.""" + try: + from litellm.proxy._experimental.mcp_server.server import ( + handle_streamable_http_mcp, + session_manager_stateless, + ) + except ImportError: + pytest.skip("MCP server not available") + + scope = _passthrough_mode_scope( + "od_server", + extra_headers=[ + (b"x-litellm-api-key", b"Bearer sk-1234"), + (b"authorization", b"Bearer upstream-token"), + ], + ) + receive = AsyncMock( + return_value={ + "type": "http.request", + "body": b'{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}', + "more_body": False, + } + ) + send = AsyncMock() + user_auth = MagicMock() + user_auth.user_id = "u1" + od_server = _build_passthrough_mode_server("od_server", MCPAuth.oauth_delegate) + + with ( + patch( + "litellm.proxy._experimental.mcp_server.server.extract_mcp_auth_context", + new_callable=AsyncMock, + return_value=(user_auth, None, ["od_server"], None, None, None), + ), + patch("litellm.proxy._experimental.mcp_server.server.set_auth_context"), + patch( + "litellm.proxy._experimental.mcp_server.server._SESSION_MANAGERS_INITIALIZED", + True, + ), + patch( + "litellm.proxy._experimental.mcp_server.server._handle_stale_mcp_session", + new_callable=AsyncMock, + return_value=False, + ), + patch( + "litellm.proxy._experimental.mcp_server.server._check_passthrough_upstream_auth", + new_callable=AsyncMock, + ), + patch( + "litellm.proxy._experimental.mcp_server.server.global_mcp_server_manager.get_mcp_server_by_name", + return_value=od_server, + ), + patch.object( + session_manager_stateless, + "handle_request", + new_callable=AsyncMock, + ) as mock_handle_request, + patch.object(session_manager_stateless, "_server_instances", {}), + ): + await handle_streamable_http_mcp(scope, receive, send) + + assert mock_handle_request.await_count == 1 + + +async def _run_passthrough_connect( + *, + auth_type, + server_names, + mcp_server_auth_headers, + scope_extra_headers=None, +): + """Drive handle_streamable_http_mcp through the preemptive-401 gate and report whether it + challenged (raised) or forwarded to the session manager. Returns (challenged, www_authenticate).""" + from litellm.proxy._experimental.mcp_server.server import ( + handle_streamable_http_mcp, + session_manager_stateless, + ) + + scope = _passthrough_mode_scope(server_names[0], extra_headers=scope_extra_headers) + receive = AsyncMock( + return_value={ + "type": "http.request", + "body": b'{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}', + "more_body": False, + } + ) + send = AsyncMock() + user_auth = MagicMock() + user_auth.user_id = "u1" + server = _build_passthrough_mode_server(server_names[0], auth_type) + + with ( + patch( + "litellm.proxy._experimental.mcp_server.server.extract_mcp_auth_context", + new_callable=AsyncMock, + return_value=(user_auth, None, server_names, mcp_server_auth_headers, None, None), + ), + patch("litellm.proxy._experimental.mcp_server.server.set_auth_context"), + patch("litellm.proxy._experimental.mcp_server.server._SESSION_MANAGERS_INITIALIZED", True), + patch( + "litellm.proxy._experimental.mcp_server.server._handle_stale_mcp_session", + new_callable=AsyncMock, + return_value=False, + ), + patch( + "litellm.proxy._experimental.mcp_server.server._check_passthrough_upstream_auth", + new_callable=AsyncMock, + ), + patch( + "litellm.proxy._experimental.mcp_server.server.global_mcp_server_manager.get_mcp_server_by_name", + return_value=server, + ), + patch.object(session_manager_stateless, "handle_request", new_callable=AsyncMock) as mock_handle_request, + patch.object(session_manager_stateless, "_server_instances", {}), + ): + try: + await handle_streamable_http_mcp(scope, receive, send) + except HTTPException as exc: + return True, (exc.headers or {}).get("www-authenticate") + return mock_handle_request.await_count == 0, None + + +@pytest.mark.asyncio +@pytest.mark.parametrize("auth_type", [MCPAuth.oauth_delegate, MCPAuth.true_passthrough]) +async def test_handle_streamable_http_mcp_per_server_header_skips_preemptive_challenge(auth_type): + """A per-server x-mcp-{alias}-authorization header binds the upstream token to one server; the + connect gate must recognize it and forward instead of spuriously 401-ing, since egress already + honors it. Without this, the mandatory multi-server binding is unusable at connect.""" + try: + from litellm.proxy._experimental.mcp_server.server import handle_streamable_http_mcp # noqa: F401 + except ImportError: + pytest.skip("MCP server not available") + + challenged, _ = await _run_passthrough_connect( + auth_type=auth_type, + server_names=["pt_server"], + mcp_server_auth_headers={"pt_server": {"Authorization": "Bearer upstream-token"}}, + ) + assert challenged is False + + +@pytest.mark.asyncio +@pytest.mark.parametrize("auth_type", [MCPAuth.oauth_delegate, MCPAuth.true_passthrough]) +async def test_handle_streamable_http_mcp_sanitized_per_server_header_skips_preemptive_challenge(auth_type): + """A dashboard client sends x-mcp-{sanitize_mcp_alias_for_header(alias)}-authorization, so the + alias 'pt-server' arrives as the header key 'pt_server'. Egress resolves that via the sanitized + alias, so the connect gate must too, or it 401s a token egress would forward.""" + try: + from litellm.proxy._experimental.mcp_server.server import handle_streamable_http_mcp # noqa: F401 + except ImportError: + pytest.skip("MCP server not available") + + challenged, _ = await _run_passthrough_connect( + auth_type=auth_type, + server_names=["pt-server"], + mcp_server_auth_headers={"pt_server": {"Authorization": "Bearer upstream-token"}}, + ) + assert challenged is False + + +@pytest.mark.asyncio +@pytest.mark.parametrize("auth_type", [MCPAuth.oauth_delegate, MCPAuth.true_passthrough]) +async def test_handle_streamable_http_mcp_aggregate_does_not_preemptively_challenge(auth_type): + """A multi-server aggregate must degrade gracefully: the preemptive 401 is single-server only, so + one server missing a token cannot 401 the whole connect (the listing absorbs per-server failures).""" + try: + from litellm.proxy._experimental.mcp_server.server import handle_streamable_http_mcp # noqa: F401 + except ImportError: + pytest.skip("MCP server not available") + + challenged, _ = await _run_passthrough_connect( + auth_type=auth_type, + server_names=["pt_server", "pt_server_2"], + mcp_server_auth_headers=None, + ) + assert challenged is False + + +@pytest.mark.asyncio +async def test_handle_streamable_http_mcp_true_passthrough_without_token_surfaces_verbatim_upstream_challenge(): + """true_passthrough is a transparent proxy: with no client Authorization the + gateway probes the upstream and surfaces its own WWW-Authenticate verbatim, + so the client discovers and authorizes against the upstream directly. Guards + against answering initialize locally with a silent 200.""" + from fastapi import HTTPException + + try: + from litellm.proxy._experimental.mcp_server.server import ( + handle_streamable_http_mcp, + session_manager_stateful, + ) + except ImportError: + pytest.skip("MCP server not available") + + upstream_challenge = 'Bearer resource_metadata="https://upstream.example.com/.well-known/oauth-protected-resource"' + probe_response = MagicMock() + probe_response.status_code = 401 + probe_response.headers = {"www-authenticate": upstream_challenge} + probe_client = MagicMock() + probe_client.post = AsyncMock(return_value=probe_response) + + scope = _passthrough_mode_scope("tp_server") + receive = AsyncMock( + return_value={ + "type": "http.request", + "body": b'{"jsonrpc":"2.0","id":1,"method":"initialize","params":{}}', + "more_body": False, + } + ) + send = AsyncMock() + user_auth = MagicMock() + user_auth.user_id = None + tp_server = _build_passthrough_mode_server("tp_server", MCPAuth.true_passthrough) + + with ( + patch( + "litellm.proxy._experimental.mcp_server.server.extract_mcp_auth_context", + new_callable=AsyncMock, + return_value=(user_auth, None, ["tp_server"], None, None, None), + ), + patch("litellm.proxy._experimental.mcp_server.server.set_auth_context"), + patch( + "litellm.proxy._experimental.mcp_server.server._SESSION_MANAGERS_INITIALIZED", + True, + ), + patch( + "litellm.proxy._experimental.mcp_server.server.get_async_httpx_client", + return_value=probe_client, + ), + patch( + "litellm.proxy._experimental.mcp_server.server.global_mcp_server_manager.get_mcp_server_by_name", + return_value=tp_server, + ), + patch.object( + session_manager_stateful, + "handle_request", + new_callable=AsyncMock, + ) as mock_handle_request, + ): + with pytest.raises(HTTPException) as exc_info: + await handle_streamable_http_mcp(scope, receive, send) + + assert mock_handle_request.await_count == 0 + assert exc_info.value.status_code == 401 + assert exc_info.value.headers["www-authenticate"] == upstream_challenge + probe_client.post.assert_awaited_once() + + +@pytest.mark.asyncio +async def test_handle_streamable_http_mcp_true_passthrough_with_token_skips_probe_and_challenge(): + """When the true_passthrough caller already carries an Authorization the + gateway must forward without probing or challenging. Guards the + ``not _scope_has_authorization_header(scope)`` condition and the no-probe + fast path.""" + try: + from litellm.proxy._experimental.mcp_server.server import ( + handle_streamable_http_mcp, + session_manager_stateless, + ) + except ImportError: + pytest.skip("MCP server not available") + + probe_client = MagicMock() + probe_client.post = AsyncMock() + + scope = _passthrough_mode_scope( + "tp_server", + extra_headers=[(b"authorization", b"Bearer upstream-token")], + ) + receive = AsyncMock( + return_value={ + "type": "http.request", + "body": b'{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}', + "more_body": False, + } + ) + send = AsyncMock() + user_auth = MagicMock() + user_auth.user_id = None + tp_server = _build_passthrough_mode_server("tp_server", MCPAuth.true_passthrough) + + with ( + patch( + "litellm.proxy._experimental.mcp_server.server.extract_mcp_auth_context", + new_callable=AsyncMock, + return_value=(user_auth, None, ["tp_server"], None, None, None), + ), + patch("litellm.proxy._experimental.mcp_server.server.set_auth_context"), + patch( + "litellm.proxy._experimental.mcp_server.server._SESSION_MANAGERS_INITIALIZED", + True, + ), + patch( + "litellm.proxy._experimental.mcp_server.server._handle_stale_mcp_session", + new_callable=AsyncMock, + return_value=False, + ), + patch( + "litellm.proxy._experimental.mcp_server.server._check_passthrough_upstream_auth", + new_callable=AsyncMock, + ), + patch( + "litellm.proxy._experimental.mcp_server.server.get_async_httpx_client", + return_value=probe_client, + ), + patch( + "litellm.proxy._experimental.mcp_server.server.global_mcp_server_manager.get_mcp_server_by_name", + return_value=tp_server, + ), + patch.object( + session_manager_stateless, + "handle_request", + new_callable=AsyncMock, + ) as mock_handle_request, + patch.object(session_manager_stateless, "_server_instances", {}), + ): + await handle_streamable_http_mcp(scope, receive, send) + + assert mock_handle_request.await_count == 1 + probe_client.post.assert_not_awaited() diff --git a/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_tool_search.py b/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_tool_search.py index 5c2a04456b0..b8f0b205831 100644 --- a/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_tool_search.py +++ b/tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_tool_search.py @@ -433,7 +433,7 @@ class TestCallToolRestApiVirtualTools: return_value=fake_result, ) as mock_execute, patch( - "litellm.proxy._experimental.mcp_server.rest_endpoints._fire_mcp_success_logging", + "litellm.proxy._experimental.mcp_server.rest_endpoints._fire_mcp_tool_call_logging", new_callable=AsyncMock, side_effect=RuntimeError("logging failed"), ) as mock_fire_logging, diff --git a/tests/test_litellm/proxy/_experimental/mcp_server/test_rest_endpoints.py b/tests/test_litellm/proxy/_experimental/mcp_server/test_rest_endpoints.py index e114f46e866..465cebcc12c 100644 --- a/tests/test_litellm/proxy/_experimental/mcp_server/test_rest_endpoints.py +++ b/tests/test_litellm/proxy/_experimental/mcp_server/test_rest_endpoints.py @@ -37,10 +37,7 @@ def _build_request( body_bytes = body else: body_bytes = b"" - raw_headers = [ - (key.lower().encode("latin-1"), value.encode("latin-1")) - for key, value in headers.items() - ] + raw_headers = [(key.lower().encode("latin-1"), value.encode("latin-1")) for key, value in headers.items()] scope = { "type": "http", "http_version": "1.1", @@ -62,25 +59,18 @@ def _build_request( def _get_route(path: str, method: str): for route in rest_endpoints.router.routes: - if getattr(route, "path", None) == path and method in getattr( - route, "methods", set() - ): + if getattr(route, "path", None) == path and method in getattr(route, "methods", set()): return route raise AssertionError(f"Route {method} {path} not found") def _route_has_dependency(route, dependency) -> bool: - if any( - getattr(dep, "dependency", None) == dependency - for dep in getattr(route, "dependencies", []) - ): + if any(getattr(dep, "dependency", None) == dependency for dep in getattr(route, "dependencies", [])): return True dependant = getattr(route, "dependant", None) if dependant is None: return False - return any( - getattr(dep, "call", None) == dependency for dep in dependant.dependencies - ) + return any(getattr(dep, "call", None) == dependency for dep in dependant.dependencies) class TestExecuteWithMcpClient: @@ -104,9 +94,7 @@ class TestExecuteWithMcpClient: auth_type=MCPAuth.none, ) - result = await rest_endpoints._execute_with_mcp_client( - payload, failing_operation - ) + result = await rest_endpoints._execute_with_mcp_client(payload, failing_operation) assert result["status"] == "error" assert "stack_trace" not in result @@ -267,15 +255,10 @@ class TestExecuteWithMcpClient: assert result["status"] == "ok" # The incoming Authorization must be dropped — extra_headers should # contain no oauth2 headers (only static_headers, which are None here). - assert ( - captured["extra_headers"] is None - or "Authorization" not in captured["extra_headers"] - ) + assert captured["extra_headers"] is None or "Authorization" not in captured["extra_headers"] @pytest.mark.asyncio - async def test_interactive_oauth_resolves_forwarded_token_via_presented_store( - self, monkeypatch - ): + async def test_interactive_oauth_resolves_forwarded_token_via_presented_store(self, monkeypatch): """Interactive authorization_code preview (oauth2, no client credentials): the forwarded just-authorized token is resolved THROUGH the v2 resolver via a one-shot presented store (cred_provider), not the caller-override path. The bare token (Bearer stripped) is the @@ -433,9 +416,7 @@ class TestExecuteWithMcpClient: return None async def fake_create_client(*args, **kwargs): - raise BaseExceptionGroup( - "test group", [RuntimeError("Cancelled via cancel scope")] - ) + raise BaseExceptionGroup("test group", [RuntimeError("Cancelled via cancel scope")]) monkeypatch.setattr( rest_endpoints.global_mcp_server_manager, @@ -497,9 +478,7 @@ class TestTestToolsList: "message": "Successfully retrieved tools", } - monkeypatch.setattr( - rest_endpoints, "_execute_with_mcp_client", fake_execute, raising=False - ) + monkeypatch.setattr(rest_endpoints, "_execute_with_mcp_client", fake_execute, raising=False) oauth_call_counter = {"count": 0} @@ -555,9 +534,7 @@ class TestTestToolsList: "message": "Successfully retrieved tools", } - monkeypatch.setattr( - rest_endpoints, "_execute_with_mcp_client", fake_execute, raising=False - ) + monkeypatch.setattr(rest_endpoints, "_execute_with_mcp_client", fake_execute, raising=False) oauth_headers = {"Authorization": "Bearer oauth"} oauth_call_counter = {"count": 0} @@ -573,7 +550,7 @@ class TestTestToolsList: raising=False, ) - request = _build_request({"authorization": "Bearer incoming"}) + request = _build_request({"authorization": "Bearer incoming", "x-litellm-api-key": "sk-admission"}) payload = NewMCPServerRequest( server_name="example", url="https://example.com", @@ -593,6 +570,101 @@ class TestTestToolsList: assert captured["oauth2_headers"] == oauth_headers assert oauth_call_counter["count"] == 1 + @pytest.mark.parametrize("auth_type", [MCPAuth.true_passthrough, MCPAuth.oauth_delegate]) + async def test_extracts_oauth2_headers_for_client_forwarded_modes(self, monkeypatch, auth_type): + """The browser-only authorize flow sends the upstream token as Authorization; the preview + must thread it through for the client-forwarded token modes so the passthrough arm can + forward it, instead of probing the upstream unauthenticated.""" + + captured: dict = {} + + async def fake_execute( + request, + operation, + mcp_auth_header=None, + oauth2_headers=None, + raw_headers=None, + ): + captured["mcp_auth_header"] = mcp_auth_header + captured["oauth2_headers"] = oauth2_headers + return { + "tools": [], + "error": None, + "message": "Successfully retrieved tools", + } + + monkeypatch.setattr(rest_endpoints, "_execute_with_mcp_client", fake_execute, raising=False) + + oauth_headers = {"Authorization": "Bearer upstream-token"} + + monkeypatch.setattr( + auth_mcp.MCPRequestHandler, + "_get_oauth2_headers_from_headers", + staticmethod(lambda headers: oauth_headers), + raising=False, + ) + + request = _build_request({"authorization": "Bearer upstream-token", "x-litellm-api-key": "sk-admission"}) + payload = NewMCPServerRequest( + server_name="example", + url="https://example.com", + auth_type=auth_type, + ) + + from litellm.proxy._types import LitellmUserRoles + + result = await rest_endpoints.test_tools_list( + request, + payload, + user_api_key_dict=UserAPIKeyAuth(user_role=LitellmUserRoles.PROXY_ADMIN), + ) + + assert result["message"] == "Successfully retrieved tools" + assert captured["mcp_auth_header"] is None + assert captured["oauth2_headers"] == oauth_headers + + @pytest.mark.parametrize("auth_type", [MCPAuth.oauth2, MCPAuth.true_passthrough, MCPAuth.oauth_delegate]) + async def test_does_not_forward_authorization_that_satisfied_admission(self, monkeypatch, auth_type): + """Authorization is also the admission fallback: with no x-litellm-api-key on the request, + the Authorization value is the caller's LiteLLM key, so forwarding it would send the + admission credential to the upstream.""" + + captured: dict = {} + + async def fake_execute( + request, + operation, + mcp_auth_header=None, + oauth2_headers=None, + raw_headers=None, + ): + captured["oauth2_headers"] = oauth2_headers + return { + "tools": [], + "error": None, + "message": "Successfully retrieved tools", + } + + monkeypatch.setattr(rest_endpoints, "_execute_with_mcp_client", fake_execute, raising=False) + + request = _build_request({"authorization": "Bearer sk-litellm-admission-key"}) + payload = NewMCPServerRequest( + server_name="example", + url="https://example.com", + auth_type=auth_type, + ) + + from litellm.proxy._types import LitellmUserRoles + + result = await rest_endpoints.test_tools_list( + request, + payload, + user_api_key_dict=UserAPIKeyAuth(user_role=LitellmUserRoles.PROXY_ADMIN), + ) + + assert result["message"] == "Successfully retrieved tools" + assert captured["oauth2_headers"] is None + class TestListToolsRestAPI: pytestmark = pytest.mark.asyncio @@ -722,9 +794,7 @@ class TestListToolsRestAPI: stub_server = StubServer() captured = {} - async def fake_get_tools( - server, server_auth_header, *args, apply_tool_filters=True, **kwargs - ): + async def fake_get_tools(server, server_auth_header, *args, apply_tool_filters=True, **kwargs): captured["apply_tool_filters"] = apply_tool_filters return ["tool-1"] @@ -772,9 +842,7 @@ class TestListToolsRestAPI: assert captured["apply_tool_filters"] is True @pytest.mark.parametrize("upstream_status", [401, 403]) - async def test_upstream_auth_failure_surfaces_status_and_challenge( - self, monkeypatch, upstream_status - ): + async def test_upstream_auth_failure_surfaces_status_and_challenge(self, monkeypatch, upstream_status): """A single-server pass-through request whose upstream rejects the token must surface the upstream status (401 or 403) plus its WWW-Authenticate challenge, not collapse into a 200 ``unexpected_error`` body.""" @@ -1362,9 +1430,7 @@ class TestListToolsRestAPI: oauth_headers = {"Authorization": "Bearer user-oauth-token"} - async def fake_get_user_oauth_extra_headers( - server, user_api_key_dict, prefetched_creds=None - ): + async def fake_get_user_oauth_extra_headers(server, user_api_key_dict, prefetched_creds=None): return oauth_headers captured = {} @@ -1559,7 +1625,7 @@ class TestCallToolRestAPI: fire_logging = AsyncMock(side_effect=RuntimeError("logging failed")) monkeypatch.setattr( rest_endpoints, - "_fire_mcp_success_logging", + "_fire_mcp_tool_call_logging", fire_logging, raising=False, ) @@ -1590,13 +1656,13 @@ class TestCallToolRestAPI: fire_logging = AsyncMock(side_effect=asyncio.CancelledError()) monkeypatch.setattr( rest_endpoints, - "_fire_mcp_success_logging", + "_fire_mcp_tool_call_logging", fire_logging, raising=False, ) with pytest.raises(asyncio.CancelledError): - await rest_endpoints._safe_fire_mcp_success_logging( + await rest_endpoints._safe_fire_mcp_tool_call_logging( object(), {"result": "ok"}, datetime.now(), datetime.now() ) @@ -1608,9 +1674,7 @@ class TestGetToolsForSingleServer: pytestmark = pytest.mark.asyncio - async def test_filters_tools_by_object_permission_mcp_tool_permissions( - self, monkeypatch - ): + async def test_filters_tools_by_object_permission_mcp_tool_permissions(self, monkeypatch): """Test that tools are filtered by user_api_key_auth.object_permission.mcp_tool_permissions""" from litellm.proxy._experimental.mcp_server.server import MCPServer from litellm.proxy._types import LiteLLM_ObjectPermissionTable @@ -1773,9 +1837,7 @@ class TestGetToolsForSingleServer: # All tools should be returned assert len(result) == 2 - async def test_no_filtering_when_server_not_in_mcp_tool_permissions( - self, monkeypatch - ): + async def test_no_filtering_when_server_not_in_mcp_tool_permissions(self, monkeypatch): """Test that all tools are returned when server is not in mcp_tool_permissions""" from litellm.proxy._experimental.mcp_server.server import MCPServer from litellm.proxy._types import LiteLLM_ObjectPermissionTable @@ -1828,9 +1890,7 @@ class TestGetToolsForSingleServer: # All tools should be returned since server is not in permissions assert len(result) == 2 - async def test_combines_server_allowed_tools_and_object_permission_filters( - self, monkeypatch - ): + async def test_combines_server_allowed_tools_and_object_permission_filters(self, monkeypatch): """Test that both server.allowed_tools and object_permission.mcp_tool_permissions filters are applied""" from litellm.proxy._experimental.mcp_server.server import MCPServer from litellm.proxy._types import LiteLLM_ObjectPermissionTable @@ -2148,9 +2208,7 @@ class TestPreviewOpenAPITools: "paths": { "/repos/{owner}/{repo}/actions/jobs/{job_id}/logs": { "get": { - "operationId": ( - "actions/download-job-logs-for-workflow-run" - ), + "operationId": ("actions/download-job-logs-for-workflow-run"), "summary": "Download job logs", } }, @@ -2193,9 +2251,7 @@ class TestPreviewOpenAPITools: names = [t["name"] for t in result["tools"]] anthropic_re = re.compile(r"^[a-zA-Z0-9_-]{1,128}$") for name in names: - assert anthropic_re.match( - name - ), f"preview tool name {name!r} violates ^[a-zA-Z0-9_-]+$" + assert anthropic_re.match(name), f"preview tool name {name!r} violates ^[a-zA-Z0-9_-]+$" assert "actions_download-job-logs-for-workflow-run" in names assert "pulls_list-files" in names @@ -2252,9 +2308,7 @@ class TestPreviewOpenAPITools: registered_summary_to_name: dict = {} - def fake_create_tool_function( - path, method, operation, base_url - ): # noqa: ANN001 + def fake_create_tool_function(path, method, operation, base_url): # noqa: ANN001 def _f(): return None @@ -2267,9 +2321,7 @@ class TestPreviewOpenAPITools: ) class _StubRegistry: - def register_tool( - self, name, description, input_schema, handler - ): # noqa: ANN001 + def register_tool(self, name, description, input_schema, handler): # noqa: ANN001 registered_summary_to_name[description] = name monkeypatch.setattr( @@ -2278,9 +2330,7 @@ class TestPreviewOpenAPITools: _StubRegistry(), ) - openapi_to_mcp_generator.register_tools_from_openapi( - spec, base_url="https://example.invalid" - ) + openapi_to_mcp_generator.register_tools_from_openapi(spec, base_url="https://example.invalid") assert preview_summary_to_name == registered_summary_to_name, ( f"preview {preview_summary_to_name} != " @@ -2308,15 +2358,11 @@ class TestConnectionErrorMessage: assert secret not in message def test_connect_error_points_at_reachability(self): - message = rest_endpoints._connection_error_message( - httpx.ConnectError("All connection attempts failed") - ) + message = rest_endpoints._connection_error_message(httpx.ConnectError("All connection attempts failed")) assert "unreachable" in message.lower() def test_timeout_error_message(self): - message = rest_endpoints._connection_error_message( - httpx.ConnectTimeout("timed out") - ) + message = rest_endpoints._connection_error_message(httpx.ConnectTimeout("timed out")) assert "unreachable" in message.lower() def test_http_status_error_includes_status_code(self): diff --git a/tests/test_litellm/proxy/auth/test_route_checks.py b/tests/test_litellm/proxy/auth/test_route_checks.py index d623149ff6a..204e6a671e3 100644 --- a/tests/test_litellm/proxy/auth/test_route_checks.py +++ b/tests/test_litellm/proxy/auth/test_route_checks.py @@ -2595,6 +2595,135 @@ def test_org_admin_of_multiple_orgs_can_operate_on_both(): assert _user_is_org_admin({"organizations": ["org-A", "org-B"]}, user_obj) is True +# ── LIT-4221: /team/update org-context resolution from team_id ──────────────── +from litellm.proxy.auth.auth_checks_organization import ( + add_team_org_context_to_request_body, +) + + +@pytest.mark.asyncio +async def test_add_team_org_context_resolves_org_from_team(): + """For /team/update with only team_id, the target team's org is resolved and + injected so the org-admin route gate can see it. This is what lets an org + admin update a team budget from the Hub UI, which sends team_id, not + organization_id (LIT-4221).""" + + async def fetch(team_id: str): + assert team_id == "team-1" + return "org-1" + + out = await add_team_org_context_to_request_body( + route="/team/update", + request_body={"team_id": "team-1", "max_budget": 42}, + fetch_team_org_id=fetch, + ) + assert out == {"team_id": "team-1", "max_budget": 42, "organization_id": "org-1"} + + +@pytest.mark.asyncio +async def test_add_team_org_context_noop_when_org_id_already_present(): + """If the caller already passed organization_id, no lookup happens and the + body is returned unchanged.""" + + async def fetch(team_id: str): + raise AssertionError("must not resolve when organization_id is present") + + body = {"team_id": "team-1", "organization_id": "org-explicit"} + out = await add_team_org_context_to_request_body( + route="/team/update", request_body=body, fetch_team_org_id=fetch + ) + assert out == body + + +@pytest.mark.asyncio +async def test_add_team_org_context_noop_for_other_routes(): + """Only /team/update opts into org resolution; other routes are untouched.""" + + async def fetch(team_id: str): + raise AssertionError("must not resolve for a non-opted-in route") + + body = {"team_id": "team-1"} + out = await add_team_org_context_to_request_body( + route="/team/delete", request_body=body, fetch_team_org_id=fetch + ) + assert out == body + + +@pytest.mark.asyncio +async def test_add_team_org_context_noop_when_team_has_no_org(): + """A standalone team (no org) resolves to None, so nothing is injected and + the org-admin branch stays unreachable (no blanket access).""" + + async def fetch(team_id: str): + return None + + body = {"team_id": "team-1"} + out = await add_team_org_context_to_request_body( + route="/team/update", request_body=body, fetch_team_org_id=fetch + ) + assert out == body + + +def test_team_update_gate_allows_org_admin_with_resolved_org(): + """Post-resolution (organization_id present), an org admin of that org clears + the gate for /team/update.""" + user_obj = _make_org_admin_user("org-1") + valid_token = UserAPIKeyAuth(user_id="org-admin-user", user_role=LitellmUserRoles.INTERNAL_USER.value) + request = MagicMock(spec=Request) + request.method = "POST" + request.query_params = {} + + RouteChecks.non_proxy_admin_allowed_routes_check( + user_obj=user_obj, + _user_role=LitellmUserRoles.INTERNAL_USER.value, + route="/team/update", + request=request, + valid_token=valid_token, + request_data={"team_id": "team-1", "organization_id": "org-1"}, + ) + + +def test_team_update_gate_rejects_without_org_context(): + """Without organization_id (i.e. resolution found no org, or a non-org-admin), + the gate still rejects /team/update — the fix adds no blanket allow. Guards + against re-widening the route (e.g. dropping it into self_managed_routes).""" + user_obj = _make_org_admin_user("org-1") + valid_token = UserAPIKeyAuth(user_id="org-admin-user", user_role=LitellmUserRoles.INTERNAL_USER.value) + request = MagicMock(spec=Request) + request.method = "POST" + request.query_params = {} + + with pytest.raises(Exception): + RouteChecks.non_proxy_admin_allowed_routes_check( + user_obj=user_obj, + _user_role=LitellmUserRoles.INTERNAL_USER.value, + route="/team/update", + request=request, + valid_token=valid_token, + request_data={"team_id": "team-1", "max_budget": 42}, + ) + + +def test_team_update_gate_rejects_cross_org_admin_with_resolved_org(): + """Even after the target team's org is resolved, an org admin of a DIFFERENT + org is rejected at the gate (no cross-org escalation).""" + user_obj = _make_org_admin_user("org-1") + valid_token = UserAPIKeyAuth(user_id="org-admin-user", user_role=LitellmUserRoles.INTERNAL_USER.value) + request = MagicMock(spec=Request) + request.method = "POST" + request.query_params = {} + + with pytest.raises(Exception): + RouteChecks.non_proxy_admin_allowed_routes_check( + user_obj=user_obj, + _user_role=LitellmUserRoles.INTERNAL_USER.value, + route="/team/update", + request=request, + valid_token=valid_token, + request_data={"team_id": "team-1", "organization_id": "org-2"}, + ) + + @pytest.mark.asyncio async def test_initialize_pass_through_registers_wildcard_for_auth_subpath(): """ diff --git a/tests/test_litellm/proxy/db/conftest.py b/tests/test_litellm/proxy/db/conftest.py new file mode 100644 index 00000000000..a0fb6bed4fa --- /dev/null +++ b/tests/test_litellm/proxy/db/conftest.py @@ -0,0 +1,65 @@ +import os +from collections.abc import Generator +from typing import Optional + +import pytest + +DB_ENV_KEYS = ( + "IAM_TOKEN_DB_AUTH", + "DATABASE_URL", + "DIRECT_URL", + "DATABASE_URL_READ_REPLICA", + "DATABASE_HOST", + "DATABASE_PORT", + "DATABASE_USER", + "DATABASE_USERNAME", + "DATABASE_NAME", + "DATABASE_SCHEMA", + "DATABASE_PASSWORD", + "DATABASE_HOST_READ_REPLICA", + "DATABASE_PORT_READ_REPLICA", + "DATABASE_USER_READ_REPLICA", + "DATABASE_USERNAME_READ_REPLICA", + "DATABASE_NAME_READ_REPLICA", + "DATABASE_SCHEMA_READ_REPLICA", + "DATABASE_PASSWORD_READ_REPLICA", +) + +_db_env_snapshot_key = pytest.StashKey[dict[str, Optional[str]]]() + + +def _db_env_snapshot() -> dict[str, Optional[str]]: + return {key: os.environ.get(key) for key in DB_ENV_KEYS} + + +@pytest.hookimpl(wrapper=True) +def pytest_runtest_setup(item: pytest.Item) -> Generator[None, None, None]: + item.stash[_db_env_snapshot_key] = _db_env_snapshot() + return (yield) + + +@pytest.hookimpl(wrapper=True) +def pytest_runtest_teardown(item: pytest.Item, nextitem: Optional[pytest.Item]) -> Generator[None, None, None]: + result = yield + before = item.stash[_db_env_snapshot_key] + leaked = {key: value for key, value in _db_env_snapshot().items() if value != before[key]} + for key, original in before.items(): + if original is None: + os.environ.pop(key, None) + else: + os.environ[key] = original + assert not leaked, ( + f"{item.nodeid} leaked DB env vars past monkeypatch teardown: {leaked}. " + "Product code under test writes DATABASE_URL(_READ_REPLICA) into os.environ as a side effect; " + "monkeypatch only restores keys it has a record for, so a value written to a previously unset " + "key survives the test and poisons every later test in this pytest-xdist worker process " + "(DB-backed e2e tests arm themselves on DATABASE_URL and then fail to connect). " + "Use the unset_database_url fixture (or monkeypatch.setenv) so restoration is registered." + ) + return result + + +@pytest.fixture +def unset_database_url(monkeypatch: pytest.MonkeyPatch) -> None: + monkeypatch.setenv("DATABASE_URL", "about-to-be-unset") + monkeypatch.delenv("DATABASE_URL") diff --git a/tests/test_litellm/proxy/db/test_db_url_settings.py b/tests/test_litellm/proxy/db/test_db_url_settings.py index 573bd5ae584..e5aa09addab 100644 --- a/tests/test_litellm/proxy/db/test_db_url_settings.py +++ b/tests/test_litellm/proxy/db/test_db_url_settings.py @@ -51,25 +51,21 @@ _MANAGED_DB_ENV_VARS = ( @pytest.fixture(autouse=True) -def _scrub_db_env(): +def _scrub_db_env(monkeypatch): """Start each test from a clean slate and restore the original env afterward. - ``apply_to_env`` writes ``DATABASE_URL`` straight into ``os.environ``, which - ``monkeypatch`` cannot undo. Snapshotting and restoring here keeps a - synthesized URL (e.g. ``writer.example.com``) from leaking into later tests - that read ``DATABASE_URL`` to decide whether to hit a real database. + ``apply_to_env`` writes ``DATABASE_URL`` straight into ``os.environ``. + Registering a setenv+delenv pair per var gives ``monkeypatch`` a restore + record even for previously unset keys, so a synthesized URL (e.g. + ``writer.example.com``) cannot leak into later tests that read + ``DATABASE_URL`` to decide whether to hit a real database. Restoring via + the same ``monkeypatch`` instance the tests use also keeps undo ordering + consistent (a hand-rolled snapshot/restore runs before ``monkeypatch``'s + own undo and gets clobbered by it). """ - saved = {var: os.environ.get(var) for var in _MANAGED_DB_ENV_VARS} for var in _MANAGED_DB_ENV_VARS: - os.environ.pop(var, None) - try: - yield - finally: - for var, value in saved.items(): - if value is None: - os.environ.pop(var, None) - else: - os.environ[var] = value + monkeypatch.setenv(var, "scrubbed") + monkeypatch.delenv(var) def _stub_iam_token(token: str = "FAKE_TOKEN"): diff --git a/tests/test_litellm/proxy/db/test_rds_iam_token_expiry.py b/tests/test_litellm/proxy/db/test_rds_iam_token_expiry.py index b92fd86ed7a..ca24f856022 100644 --- a/tests/test_litellm/proxy/db/test_rds_iam_token_expiry.py +++ b/tests/test_litellm/proxy/db/test_rds_iam_token_expiry.py @@ -26,25 +26,13 @@ class TestPrismaWrapperTokenRefresh: """Tests for the PrismaWrapper RDS IAM token refresh implementation.""" @pytest.fixture - def setup_env(self): + def setup_env(self, monkeypatch, unset_database_url): """Setup environment variables for testing.""" - os.environ["DATABASE_HOST"] = "test-host.rds.amazonaws.com" - os.environ["DATABASE_PORT"] = "5432" - os.environ["DATABASE_USER"] = "test_user" - os.environ["DATABASE_NAME"] = "test_db" - os.environ["IAM_TOKEN_DB_AUTH"] = "True" - yield - # Cleanup - for key in [ - "DATABASE_HOST", - "DATABASE_PORT", - "DATABASE_USER", - "DATABASE_NAME", - "DATABASE_URL", - "IAM_TOKEN_DB_AUTH", - "DATABASE_SCHEMA", - ]: - os.environ.pop(key, None) + monkeypatch.setenv("DATABASE_HOST", "test-host.rds.amazonaws.com") + monkeypatch.setenv("DATABASE_PORT", "5432") + monkeypatch.setenv("DATABASE_USER", "test_user") + monkeypatch.setenv("DATABASE_NAME", "test_db") + monkeypatch.setenv("IAM_TOKEN_DB_AUTH", "True") def _generate_mock_token(self, expires_in_seconds: int = 900) -> str: """Generate a mock IAM token with expiration info.""" @@ -172,22 +160,12 @@ class TestBackgroundRefreshLoop: """Tests for the background refresh loop timing.""" @pytest.fixture - def setup_env(self): + def setup_env(self, monkeypatch, unset_database_url): """Setup environment variables for testing.""" - os.environ["DATABASE_HOST"] = "test-host.rds.amazonaws.com" - os.environ["DATABASE_PORT"] = "5432" - os.environ["DATABASE_USER"] = "test_user" - os.environ["DATABASE_NAME"] = "test_db" - yield - # Cleanup - for key in [ - "DATABASE_HOST", - "DATABASE_PORT", - "DATABASE_USER", - "DATABASE_NAME", - "DATABASE_URL", - ]: - os.environ.pop(key, None) + monkeypatch.setenv("DATABASE_HOST", "test-host.rds.amazonaws.com") + monkeypatch.setenv("DATABASE_PORT", "5432") + monkeypatch.setenv("DATABASE_USER", "test_user") + monkeypatch.setenv("DATABASE_NAME", "test_db") @pytest.mark.asyncio async def test_calculate_seconds_fallback_when_no_url(self, setup_env): diff --git a/tests/test_litellm/proxy/db/test_routing_prisma_wrapper.py b/tests/test_litellm/proxy/db/test_routing_prisma_wrapper.py index 92043f44ca9..e5bb8b99507 100644 --- a/tests/test_litellm/proxy/db/test_routing_prisma_wrapper.py +++ b/tests/test_litellm/proxy/db/test_routing_prisma_wrapper.py @@ -626,7 +626,7 @@ async def test_getattr_does_not_block_inside_running_loop_on_expired_token(monke assert refresh_calls["count"] == 1 -def test_writer_get_rds_iam_token_defaults_port_when_unset(monkeypatch): +def test_writer_get_rds_iam_token_defaults_port_when_unset(monkeypatch, unset_database_url): """When DATABASE_PORT is unset, the writer must default to the Postgres standard port instead of passing `None` through. Passing None to `generate_iam_auth_token` makes botocore embed the literal string @@ -639,7 +639,6 @@ def test_writer_get_rds_iam_token_defaults_port_when_unset(monkeypatch): monkeypatch.setenv("DATABASE_USER", "litellm") monkeypatch.setenv("DATABASE_NAME", "litellm") monkeypatch.delenv("DATABASE_SCHEMA", raising=False) - monkeypatch.delenv("DATABASE_URL", raising=False) captured: Dict[str, Any] = {} @@ -661,7 +660,7 @@ def test_writer_get_rds_iam_token_defaults_port_when_unset(monkeypatch): assert ":5432/litellm" in (new_url or "") -def test_writer_get_rds_iam_token_uses_database_host_env_vars(monkeypatch): +def test_writer_get_rds_iam_token_uses_database_host_env_vars(monkeypatch, unset_database_url): """Writer's IAM path (no iam_endpoint configured) reads host/port/user/db from the legacy DATABASE_HOST/PORT/USER/NAME env vars and writes the URL back to DATABASE_URL — this is the pre-read-replica behavior the patch @@ -673,7 +672,6 @@ def test_writer_get_rds_iam_token_uses_database_host_env_vars(monkeypatch): monkeypatch.setenv("DATABASE_USER", "litellm") monkeypatch.setenv("DATABASE_NAME", "litellm") monkeypatch.setenv("DATABASE_SCHEMA", "public") - monkeypatch.delenv("DATABASE_URL", raising=False) captured: Dict[str, Any] = {} diff --git a/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_aim.py b/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_aim.py new file mode 100644 index 00000000000..2e83422074e --- /dev/null +++ b/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_aim.py @@ -0,0 +1,88 @@ +"""Tests for the AIM guardrail's inspection-payload construction.""" + +from litellm.proxy.guardrails.guardrail_hooks.aim.aim import AimGuardrail + + +def test_aim_inspection_messages_coerces_chat_completions_tool_role_to_user(): + """LIT-4294: A valid chat-completions ``role: "tool"`` message carries a + ``tool_call_id``, but the inspection flatten drops every field except + ``role`` and ``content``. A bare ``tool`` message without ``tool_call_id`` + is schema-invalid per the OpenAI chat schema, and the customer's writeup + reproduced AIM's ``/fw/v1/analyze`` returning 422 on exactly that shape. + The AIM POST collapses the role to ``user``; the outbound request to the + LLM is untouched.""" + data = { + "messages": [ + {"role": "user", "content": "weather in SF"}, + { + "role": "assistant", + "content": "", + "tool_calls": [ + { + "id": "c1", + "type": "function", + "function": {"name": "get_weather", "arguments": "{}"}, + } + ], + }, + {"role": "tool", "tool_call_id": "c1", "content": "sunny"}, + ] + } + assert AimGuardrail._build_aim_inspection_messages(data) == [ + {"role": "user", "content": "weather in SF"}, + {"role": "user", "content": "sunny"}, + ] + + +def test_aim_inspection_messages_coerces_non_standard_caller_role_to_user(): + """LIT-4294: A caller-supplied role outside {system, user, assistant} + (e.g. ``developer``, ``function``) is coerced to ``user`` for the AIM + POST, since AIM validates the payload against the OpenAI chat schema + and rejects unknown roles the same way it rejects bare ``tool``.""" + data = { + "messages": [ + {"role": "developer", "content": "system-ish instruction"}, + {"role": "user", "content": "normal user text"}, + ] + } + assert AimGuardrail._build_aim_inspection_messages(data) == [ + {"role": "user", "content": "system-ish instruction"}, + {"role": "user", "content": "normal user text"}, + ] + + +def test_aim_inspection_messages_coerces_responses_function_call_output_role(): + """LIT-4294: the shared helper synthesises ``role: "tool"`` for a + Responses ``function_call_output`` item (semantic equivalent of + chat-completions tool messages). AIM's schema-validating POST cannot + carry ``tool_call_id`` in the flat inspection payload, so AIM collapses + that ``tool`` role to ``user`` locally before POSTing.""" + data = { + "input": [ + { + "type": "function_call_output", + "call_id": "c1", + "output": [{"type": "input_text", "text": "sunny"}], + }, + ] + } + assert AimGuardrail._build_aim_inspection_messages(data) == [ + {"role": "user", "content": "sunny"}, + ] + + +def test_aim_inspection_messages_preserves_safe_roles(): + """Safe roles pass through untouched — the coercion only fires for + roles the OpenAI chat schema flatten cannot represent standalone.""" + data = { + "messages": [ + {"role": "system", "content": "be helpful"}, + {"role": "user", "content": "hi"}, + {"role": "assistant", "content": "hello"}, + ] + } + assert AimGuardrail._build_aim_inspection_messages(data) == [ + {"role": "system", "content": "be helpful"}, + {"role": "user", "content": "hi"}, + {"role": "assistant", "content": "hello"}, + ] diff --git a/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_bedrock_guardrails.py b/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_bedrock_guardrails.py index f43d8e85aca..15827b80bcf 100644 --- a/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_bedrock_guardrails.py +++ b/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_bedrock_guardrails.py @@ -2767,3 +2767,510 @@ async def test_grounding_output_blocked_raises_400(): ) assert exc_info.value.status_code == 400 + + +############################################################################### +# LIT-4186: disable_exception_on_block regression tests +# +# Before the fix, a Bedrock block with disable_exception_on_block=True raised +# GuardrailInterventionNormalStringError, which no proxy code handled: the +# unified pre_call path re-raised it, so the client saw HTTP 500 with the block +# message; the native during_call hook swallowed it and set data["mock_response"], +# which was dead code because route_request already unpacked kwargs. +# +# The fix converts blocks to ModifyResponseException at the raise site inside +# make_bedrock_api_request. That exception is already the industry-standard +# proxy contract (caught in proxy_server.py, anthropic_endpoints, etc.) and +# turns into a 200 response whose content is the block message. +############################################################################### + + +def _blocked_bedrock_httpx_response() -> MagicMock: + response = MagicMock() + response.status_code = 200 + response.json.return_value = { + "action": "GUARDRAIL_INTERVENED", + "outputs": [{"text": "Sorry, the model cannot answer this question."}], + "assessments": [ + { + "topicPolicy": { + "topics": [{"name": "Denied", "type": "DENY", "action": "BLOCKED"}] + } + } + ], + } + return response + + +@pytest.mark.asyncio +async def test_make_bedrock_api_request_block_raises_modify_response_when_flag_set(): + from litellm.exceptions import ModifyResponseException + + guardrail = BedrockGuardrail( + guardrail_name="test-bedrock-guard", + guardrailIdentifier="test-guardrail", + guardrailVersion="DRAFT", + disable_exception_on_block=True, + ) + + request_data = {"model": "bedrock-nova-micro"} + mock_credentials = MagicMock() + mock_credentials.access_key = "k" + mock_credentials.secret_key = "s" + mock_credentials.token = None + + with ( + patch.object( + guardrail.async_handler, "post", new_callable=AsyncMock + ) as mock_post, + patch.object( + guardrail, "_load_credentials", return_value=(mock_credentials, "us-east-1") + ), + patch.object(guardrail, "_prepare_request", return_value=MagicMock()), + ): + mock_post.return_value = _blocked_bedrock_httpx_response() + + with pytest.raises(ModifyResponseException) as exc_info: + await guardrail.make_bedrock_api_request( + source="INPUT", + messages=[{"role": "user", "content": "My name is John Doe"}], + request_data=request_data, + ) + + assert exc_info.value.message == "Sorry, the model cannot answer this question." + assert exc_info.value.model == "bedrock-nova-micro" + assert exc_info.value.guardrail_name == "test-bedrock-guard" + + +@pytest.mark.asyncio +async def test_make_bedrock_api_request_block_raises_http_400_when_flag_unset(): + guardrail = BedrockGuardrail( + guardrail_name="test-bedrock-guard", + guardrailIdentifier="test-guardrail", + guardrailVersion="DRAFT", + disable_exception_on_block=False, + ) + + mock_credentials = MagicMock() + mock_credentials.access_key = "k" + mock_credentials.secret_key = "s" + mock_credentials.token = None + + with ( + patch.object( + guardrail.async_handler, "post", new_callable=AsyncMock + ) as mock_post, + patch.object( + guardrail, "_load_credentials", return_value=(mock_credentials, "us-east-1") + ), + patch.object(guardrail, "_prepare_request", return_value=MagicMock()), + ): + mock_post.return_value = _blocked_bedrock_httpx_response() + + with pytest.raises(HTTPException) as exc_info: + await guardrail.make_bedrock_api_request( + source="INPUT", + messages=[{"role": "user", "content": "hi"}], + request_data={"model": "bedrock-nova-micro"}, + ) + + assert exc_info.value.status_code == 400 + + +@pytest.mark.asyncio +async def test_async_pre_call_hook_propagates_modify_response_on_block(): + """pre_call: block with disable_exception_on_block=True must raise + ModifyResponseException so the endpoint handler returns 200 with the block + message. Before LIT-4186 the exception was swallowed and only data + ["mock_response"] was mutated, which the unified pre_call path never read + (surfaced as HTTP 500).""" + from litellm.exceptions import ModifyResponseException + + guardrail = BedrockGuardrail( + guardrail_name="test-bedrock-guard", + guardrailIdentifier="test-guardrail", + guardrailVersion="DRAFT", + disable_exception_on_block=True, + ) + + request_data = { + "model": "bedrock-nova-micro", + "messages": [{"role": "user", "content": "My name is John Doe"}], + } + mock_credentials = MagicMock() + mock_credentials.access_key = "k" + mock_credentials.secret_key = "s" + mock_credentials.token = None + + with ( + patch.object( + guardrail.async_handler, "post", new_callable=AsyncMock + ) as mock_post, + patch.object( + guardrail, "_load_credentials", return_value=(mock_credentials, "us-east-1") + ), + patch.object(guardrail, "_prepare_request", return_value=MagicMock()), + ): + mock_post.return_value = _blocked_bedrock_httpx_response() + + with pytest.raises(ModifyResponseException) as exc_info: + await guardrail.async_pre_call_hook( + user_api_key_dict=UserAPIKeyAuth(), + cache=DualCache(), + data=request_data, + call_type="acompletion", + ) + + assert exc_info.value.message == "Sorry, the model cannot answer this question." + # No `mock_response` mutation: the old broken contract must be gone + # (route_request unpacks kwargs before this hook runs, so `mock_response` + # would never reach the LLM call anyway). + assert "mock_response" not in request_data + + +@pytest.mark.asyncio +async def test_async_moderation_hook_propagates_modify_response_on_block(): + """during_call: block must raise ModifyResponseException from the moderation + task so the surrounding asyncio.gather cancels the LLM call, instead of + the old behavior of swallowing the block and letting the model call proceed + (LIT-4186 symptom 2: silent bypass, model billed).""" + from litellm.exceptions import ModifyResponseException + + guardrail = BedrockGuardrail( + guardrail_name="test-bedrock-guard", + guardrailIdentifier="test-guardrail", + guardrailVersion="DRAFT", + disable_exception_on_block=True, + ) + + request_data = { + "model": "bedrock-nova-micro", + "messages": [{"role": "user", "content": "My name is John Doe"}], + } + mock_credentials = MagicMock() + mock_credentials.access_key = "k" + mock_credentials.secret_key = "s" + mock_credentials.token = None + + with ( + patch.object( + guardrail.async_handler, "post", new_callable=AsyncMock + ) as mock_post, + patch.object( + guardrail, "_load_credentials", return_value=(mock_credentials, "us-east-1") + ), + patch.object(guardrail, "_prepare_request", return_value=MagicMock()), + ): + mock_post.return_value = _blocked_bedrock_httpx_response() + + with pytest.raises(ModifyResponseException) as exc_info: + await guardrail.async_moderation_hook( + data=request_data, + user_api_key_dict=UserAPIKeyAuth(), + call_type="acompletion", + ) + + assert exc_info.value.message == "Sorry, the model cannot answer this question." + + +@pytest.mark.asyncio +async def test_async_post_call_success_hook_attaches_original_response_on_block(): + """post_call: block must raise ModifyResponseException and attach the LLM + response to `original_response` so the synthetic block reply reports the + upstream call's real token usage instead of zero.""" + from litellm.exceptions import ModifyResponseException + + guardrail = BedrockGuardrail( + guardrail_name="test-bedrock-guard", + guardrailIdentifier="test-guardrail", + guardrailVersion="DRAFT", + disable_exception_on_block=True, + ) + + request_data = { + "model": "bedrock-nova-micro", + "messages": [{"role": "user", "content": "hi"}], + } + llm_response = _model_response("Hello John Doe! The capital of France is Paris.") + mock_credentials = MagicMock() + mock_credentials.access_key = "k" + mock_credentials.secret_key = "s" + mock_credentials.token = None + + with ( + patch.object( + guardrail.async_handler, "post", new_callable=AsyncMock + ) as mock_post, + patch.object( + guardrail, "_load_credentials", return_value=(mock_credentials, "us-east-1") + ), + patch.object(guardrail, "_prepare_request", return_value=MagicMock()), + ): + mock_post.return_value = _blocked_bedrock_httpx_response() + + with pytest.raises(ModifyResponseException) as exc_info: + await guardrail.async_post_call_success_hook( + data=request_data, + user_api_key_dict=UserAPIKeyAuth(), + response=llm_response, + ) + + assert exc_info.value.original_response is llm_response + + +@pytest.mark.asyncio +async def test_apply_guardrail_propagates_modify_response_on_block(): + """apply_guardrail (unified path used by pre_call / /apply_guardrail + endpoint) must let ModifyResponseException propagate as-is so the endpoint + handler catches it and returns a 200.""" + from litellm.exceptions import ModifyResponseException + + guardrail = BedrockGuardrail( + guardrail_name="test-bedrock-guard", + guardrailIdentifier="test-guardrail", + guardrailVersion="DRAFT", + disable_exception_on_block=True, + ) + + with patch.object( + guardrail, "make_bedrock_api_request", new_callable=AsyncMock + ) as mock_api: + mock_api.side_effect = ModifyResponseException( + message="Sorry, the model cannot answer this question.", + model="bedrock-nova-micro", + request_data={}, + guardrail_name="test-bedrock-guard", + ) + + with pytest.raises(ModifyResponseException) as exc_info: + await guardrail.apply_guardrail( + inputs={"texts": ["My name is John Doe"]}, + request_data={"model": "bedrock-nova-micro"}, + input_type="request", + ) + + assert exc_info.value.message == "Sorry, the model cannot answer this question." + + +@pytest.mark.asyncio +async def test_streaming_post_call_block_yields_synthetic_stream_not_raise(): + """LIT-4186 regression: with disable_exception_on_block=True, streaming + post_call blocks must be delivered as a synthetic stream (finish_reason= + content_filter, block message as content), NOT raised. Pre-fix the local + handler already produced this shape; the LIT-4186 refactor briefly turned + it into an SSE 500 by letting ModifyResponseException escape the streaming + generator. This test locks in the correct streaming contract. + """ + from litellm.types.utils import Delta, ModelResponseStream, StreamingChoices + + guardrail = BedrockGuardrail( + guardrail_name="test-bedrock-guard", + guardrailIdentifier="test-guardrail", + guardrailVersion="DRAFT", + disable_exception_on_block=True, + ) + + async def _stream(): + yield ModelResponseStream( + choices=[ + StreamingChoices( + index=0, + delta=Delta(role="assistant", content="Coffee is a popular"), + ) + ] + ) + yield ModelResponseStream( + choices=[StreamingChoices(index=0, delta=Delta(content=" beverage."), finish_reason="stop")] + ) + + mock_credentials = MagicMock() + mock_credentials.access_key = "k" + mock_credentials.secret_key = "s" + mock_credentials.token = None + + with ( + patch.object(guardrail.async_handler, "post", new_callable=AsyncMock) as mock_post, + patch.object(guardrail, "_load_credentials", return_value=(mock_credentials, "us-east-1")), + patch.object(guardrail, "_prepare_request", return_value=MagicMock()), + ): + mock_post.return_value = _blocked_bedrock_httpx_response() + + chunks = [ + c + async for c in guardrail.async_post_call_streaming_iterator_hook( + user_api_key_dict=UserAPIKeyAuth(), + response=_stream(), + request_data={"model": "bedrock-nova-micro"}, + ) + ] + + assert chunks, "streaming block should yield synthetic chunks, not error out" + assembled_content = "".join( + (c.choices[0].delta.content or "") + for c in chunks + if getattr(c, "choices", None) and getattr(c.choices[0], "delta", None) + ) + assert assembled_content == "Sorry, the model cannot answer this question." + assert chunks[-1].choices[0].finish_reason == "content_filter" + + +@pytest.mark.asyncio +async def test_streaming_post_call_block_preserves_upstream_usage(): + """LIT-4186: streaming block must report the usage the upstream LLM call + actually consumed. Non-streaming blocks carry it via original_response + + _blocked_response_usage in the endpoint handler; streaming has to copy it + onto the synthetic ModelResponse directly since the exception can't escape + the SSE generator. Without this, clients see accurate billing on + non-streaming blocks and zero on streaming blocks -- silent revenue leak.""" + from litellm.types.utils import Delta, ModelResponseStream, StreamingChoices, Usage + + guardrail = BedrockGuardrail( + guardrail_name="test-bedrock-guard", + guardrailIdentifier="test-guardrail", + guardrailVersion="DRAFT", + disable_exception_on_block=True, + ) + + async def _stream_with_usage(): + # Terminal chunk carrying usage, as OpenAI-style streams do with + # stream_options={"include_usage": True}. stream_chunk_builder + # aggregates this into the assembled ModelResponse's .usage. + yield ModelResponseStream( + choices=[StreamingChoices(index=0, delta=Delta(role="assistant", content="Coffee is delicious"))] + ) + yield ModelResponseStream( + choices=[StreamingChoices(index=0, delta=Delta(content=""), finish_reason="stop")], + usage=Usage(prompt_tokens=42, completion_tokens=17, total_tokens=59), + ) + + mock_credentials = MagicMock() + mock_credentials.access_key = "k" + mock_credentials.secret_key = "s" + mock_credentials.token = None + + with ( + patch.object(guardrail.async_handler, "post", new_callable=AsyncMock) as mock_post, + patch.object(guardrail, "_load_credentials", return_value=(mock_credentials, "us-east-1")), + patch.object(guardrail, "_prepare_request", return_value=MagicMock()), + ): + mock_post.return_value = _blocked_bedrock_httpx_response() + + chunks = [ + c + async for c in guardrail.async_post_call_streaming_iterator_hook( + user_api_key_dict=UserAPIKeyAuth(), + response=_stream_with_usage(), + request_data={"model": "bedrock-nova-micro"}, + ) + ] + + # Find the chunk carrying usage (MockResponseIterator emits it on the + # terminating chunk when the source ModelResponse has .usage set) + usage_chunks = [c for c in chunks if getattr(c, "usage", None) is not None] + assert usage_chunks, "streaming block should carry the upstream call's usage on at least one chunk" + reported_usage = usage_chunks[-1].usage + assert reported_usage.prompt_tokens == 42 + assert reported_usage.completion_tokens == 17 + assert reported_usage.total_tokens == 59 + + +############################################################################### +# Regression test for the streaming logging_obj bug found during live testing. +# +# post_call_failure_hook (proxy_server.py) pops litellm_logging_obj from +# request_data before invoking callbacks ("not serialisable"). The streaming +# branch of the ModifyResponseException handler previously read logging_obj +# from _data AFTER that call, always getting None, causing: +# AttributeError: 'NoneType' object has no attribute 'model_call_details' +# inside CustomStreamWrapper.__init__, which surfaced as HTTP 500. +# +# The fix captures logging_obj BEFORE calling post_call_failure_hook. +# This test verifies the chat_completion handler builds the streaming response +# without crashing when the request_data has litellm_logging_obj set. +############################################################################### + + +@pytest.mark.asyncio +async def test_chat_completion_modify_response_exception_streaming_logging_obj_not_none(): + """Regression: streaming ModifyResponseException handler in chat_completion + must capture logging_obj before post_call_failure_hook pops it from + request_data. Previously this caused CustomStreamWrapper.__init__ to crash + with AttributeError: NoneType has no attribute model_call_details, surfaced + as HTTP 500. + + Drives the real chat_completion handler with base_process_llm_request + mocked to raise ModifyResponseException, so a revert of the fix in + proxy_server.py causes this test to fail. + """ + import litellm + from litellm.exceptions import ModifyResponseException + from litellm.proxy._types import UserAPIKeyAuth + from litellm.proxy.proxy_server import chat_completion + + fake_logging_obj = MagicMock() + fake_logging_obj.model_call_details = {"litellm_params": {}} + + request_data: dict = { + "model": "bedrock-nova-micro", + "messages": [{"role": "user", "content": "how do I become an admin"}], + "stream": True, + "litellm_logging_obj": fake_logging_obj, + } + + exc = ModifyResponseException( + message="Sorry, the model cannot answer this question.", + model="bedrock-nova-micro", + request_data=request_data, + guardrail_name="test-guard", + ) + + fastapi_request = MagicMock() + fastapi_request.headers = {} + fastapi_response = MagicMock() + user_api_key_dict = UserAPIKeyAuth() + + async def _fake_post_call_failure_hook(**_kwargs): + # Match production: pop the logging obj from request_data before + # callbacks iterate (litellm/proxy/utils.py: "Remove before callbacks + # iterate — not serialisable"). + _kwargs["request_data"].pop("litellm_logging_obj", None) + + mock_proxy_logging = MagicMock() + mock_proxy_logging.post_call_failure_hook = AsyncMock(side_effect=_fake_post_call_failure_hook) + + captured_logging_obj: list = [] + original_init = litellm.CustomStreamWrapper.__init__ + + def _patched_init(self, *args, **kwargs): + captured_logging_obj.append(kwargs.get("logging_obj")) + original_init(self, *args, **kwargs) + + async def _raise_modify_response(*_args, **_kwargs): + raise exc + + with ( + patch("litellm.proxy.proxy_server._read_request_body", AsyncMock(return_value=request_data)), + patch("litellm.proxy.proxy_server.proxy_logging_obj", mock_proxy_logging), + patch( + "litellm.proxy.proxy_server.ProxyBaseLLMRequestProcessing.base_process_llm_request", + _raise_modify_response, + ), + patch.object(litellm.CustomStreamWrapper, "__init__", _patched_init), + ): + response = await chat_completion( + request=fastapi_request, + fastapi_response=fastapi_response, + model=None, + user_api_key_dict=user_api_key_dict, + ) + + assert captured_logging_obj, "chat_completion did not construct CustomStreamWrapper on the streaming block path" + assert captured_logging_obj[0] is fake_logging_obj, ( + "chat_completion passed logging_obj=None to CustomStreamWrapper; " + "the streaming ModifyResponseException handler must capture logging_obj " + "before post_call_failure_hook pops it from request_data" + ) + # A streaming block returns a StreamingResponse; if the fix were reverted, + # CustomStreamWrapper would raise AttributeError inside __init__ and this + # call would never reach here. + assert response is not None diff --git a/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_grayswan.py b/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_grayswan.py index f2e7447239f..53af7f36a5f 100644 --- a/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_grayswan.py +++ b/tests/test_litellm/proxy/guardrails/guardrail_hooks/test_grayswan.py @@ -5,6 +5,7 @@ from fastapi import HTTPException from litellm.integrations.custom_guardrail import ModifyResponseException from litellm.proxy._types import UserAPIKeyAuth +from litellm.proxy.guardrails.guardrail_hooks.grayswan import grayswan as grayswan_module from litellm.proxy.guardrails.guardrail_hooks.grayswan.grayswan import ( GraySwanGuardrail, GraySwanGuardrailAPIError, @@ -70,12 +71,118 @@ def test_prepare_payload_includes_dynamic_metadata( assert payload["metadata"] == dynamic_body["metadata"] +def test_prepare_payload_forwards_only_scan_id_header( + grayswan_guardrail: GraySwanGuardrail, +) -> None: + messages = [{"role": "user", "content": "hello"}] + request_data = { + "proxy_server_request": { + "headers": { + "SHADE_SCAN_ID": "scan-123", + "authorization": "Bearer secret", + } + }, + "litellm_metadata": {"request_id": "request-123"}, + } + + payload = grayswan_guardrail._prepare_payload(messages, {}, request_data) + + assert payload["litellm_metadata"] == { + "request_id": "request-123", + "headers": {"SHADE_SCAN_ID": "scan-123"}, + } + + +def test_prepare_payload_merges_scan_id_with_existing_metadata_headers( + grayswan_guardrail: GraySwanGuardrail, +) -> None: + messages = [{"role": "user", "content": "hello"}] + request_data = { + "proxy_server_request": { + "headers": { + "shade_scan_id": "scan-123", + } + }, + "litellm_metadata": { + "request_id": "request-123", + "headers": {"x-existing": "keep-me"}, + }, + } + + payload = grayswan_guardrail._prepare_payload(messages, {}, request_data) + + assert payload["litellm_metadata"] == { + "request_id": "request-123", + "headers": { + "x-existing": "keep-me", + "shade_scan_id": "scan-123", + }, + } + + +def test_prepare_payload_sanitizes_headers_when_litellm_metadata_absent( + monkeypatch: pytest.MonkeyPatch, + grayswan_guardrail: GraySwanGuardrail, +) -> None: + messages = [{"role": "user", "content": "hello"}] + request_data = { + "proxy_server_request": { + "headers": { + "shade_scan_id": "scan-123", + } + } + } + + monkeypatch.setattr(grayswan_module, "safe_dumps", lambda _data: "{}") + + payload = grayswan_guardrail._prepare_payload(messages, {}, request_data) + + assert "litellm_metadata" not in payload + + +def test_prepare_payload_extracts_headers_from_logging_obj( + grayswan_guardrail: GraySwanGuardrail, +) -> None: + messages = [{"role": "user", "content": "hello"}] + request_data = {} + logging_obj = type( + "LoggingObj", + (), + { + "model_call_details": { + "litellm_params": { + "metadata": { + "headers": { + "shade_scan_id": "scan-from-logging", + "authorization": "Bearer secret", + } + } + } + } + }, + )() + + payload = grayswan_guardrail._prepare_payload(messages, {}, request_data, logging_obj) + + assert payload["litellm_metadata"] == { + "headers": {"shade_scan_id": "scan-from-logging"}, + } + + +def test_prepare_payload_ignores_logging_obj_without_model_call_details( + grayswan_guardrail: GraySwanGuardrail, +) -> None: + messages = [{"role": "user", "content": "hello"}] + + payload = grayswan_guardrail._prepare_payload(messages, {}, {}, object()) + + assert "litellm_metadata" not in payload + + def test_process_response_does_not_block_under_threshold( grayswan_guardrail: GraySwanGuardrail, ) -> None: - grayswan_guardrail._process_grayswan_response( - {"violation": 0.3, "violated_rules": []} - ) + grayswan_guardrail._process_grayswan_response({"violation": 0.3, "violated_rules": []}) def test_process_response_blocks_when_threshold_exceeded() -> None: @@ -127,16 +234,12 @@ class _DummyClient: self.calls: list[dict] = [] async def post(self, *, url: str, headers: dict, json: dict, timeout: float): - self.calls.append( - {"url": url, "headers": headers, "json": json, "timeout": timeout} - ) + self.calls.append({"url": url, "headers": headers, "json": json, "timeout": timeout}) return _DummyResponse(self.payload) @pytest.mark.asyncio -async def test_run_guardrail_posts_payload( - monkeypatch, grayswan_guardrail: GraySwanGuardrail -) -> None: +async def test_run_guardrail_posts_payload(monkeypatch, grayswan_guardrail: GraySwanGuardrail) -> None: dummy_client = _DummyClient({"violation": 0.1}) grayswan_guardrail.async_handler = dummy_client @@ -308,9 +411,7 @@ def test_process_response_passthrough_raises_exception_in_pre_call() -> None: # Should raise ModifyResponseException with pytest.raises(ModifyResponseException) as exc: - guardrail._process_grayswan_response( - response_json, data, GuardrailEventHooks.pre_call - ) + guardrail._process_grayswan_response(response_json, data, GuardrailEventHooks.pre_call) assert "Gray Swan Cygnal Guardrail" in exc.value.message assert exc.value.model == "gpt-4" @@ -338,9 +439,7 @@ def test_process_response_passthrough_raises_exception_in_during_call() -> None: # Should raise ModifyResponseException with pytest.raises(ModifyResponseException) as exc: - guardrail._process_grayswan_response( - response_json, data, GuardrailEventHooks.during_call - ) + guardrail._process_grayswan_response(response_json, data, GuardrailEventHooks.during_call) assert "Gray Swan Cygnal Guardrail" in exc.value.message assert exc.value.model == "gpt-4" @@ -365,9 +464,7 @@ def test_process_response_passthrough_stores_detection_info_in_post_call() -> No } # Should NOT raise an exception in post_call - guardrail._process_grayswan_response( - response_json, data, GuardrailEventHooks.post_call - ) + guardrail._process_grayswan_response(response_json, data, GuardrailEventHooks.post_call) # Verify detection info was stored in metadata assert "metadata" in data @@ -400,9 +497,7 @@ def test_process_response_passthrough_does_not_raise_if_under_threshold() -> Non } # Should not raise an exception since under threshold - guardrail._process_grayswan_response( - response_json, data, GuardrailEventHooks.pre_call - ) + guardrail._process_grayswan_response(response_json, data, GuardrailEventHooks.pre_call) # Should not have any detection info since it didn't exceed threshold assert "guardrail_detections" not in data.get("metadata", {}) @@ -436,10 +531,7 @@ def test_format_violation_message() -> None: assert "Gray Swan Cygnal Guardrail" in message assert "the input query has a violation score of 0.85" in message assert "violating the rule(s): 1, 3, 5" in message - assert ( - "Mutation effort to make the harmful intention disguised was DETECTED" - in message - ) + assert "Mutation effort to make the harmful intention disguised was DETECTED" in message # IPI should not be in message since it's False assert "Indirect Prompt Injection was DETECTED" not in message @@ -450,10 +542,7 @@ def test_format_violation_message() -> None: assert "Gray Swan Cygnal Guardrail" in message assert "the model response has a violation score of 0.85" in message assert "violating the rule(s): 1, 3, 5" in message - assert ( - "Mutation effort to make the harmful intention disguised was DETECTED" - in message - ) + assert "Mutation effort to make the harmful intention disguised was DETECTED" in message def test_prepare_payload_includes_litellm_metadata( diff --git a/tests/test_litellm/proxy/guardrails/test_content_utils.py b/tests/test_litellm/proxy/guardrails/test_content_utils.py index 099fca78a62..34d92505359 100644 --- a/tests/test_litellm/proxy/guardrails/test_content_utils.py +++ b/tests/test_litellm/proxy/guardrails/test_content_utils.py @@ -8,7 +8,6 @@ from litellm.proxy.guardrails._content_utils import ( walk_user_text, ) - # ── iter_message_text ──────────────────────────────────────────────────────────── @@ -101,6 +100,55 @@ def test_iter_message_text_empty_data(): assert list(iter_message_text({"input": ""})) == [] +def test_iter_message_text_responses_api_input_text_and_output_text_parts(): + """LIT-4294: Responses-API content parts use ``input_text`` (request) and + ``output_text`` (assistant); reading only ``type == "text"`` skipped every + ``/v1/responses`` body and every text guardrail was a no-op on that path.""" + data = { + "input": [ + { + "type": "message", + "role": "user", + "content": [{"type": "input_text", "text": "user text"}], + }, + { + "type": "message", + "role": "assistant", + "content": [{"type": "output_text", "text": "assistant text"}], + }, + ] + } + assert list(iter_message_text(data)) == ["user text", "assistant text"] + + +def test_iter_message_text_responses_api_tool_call_taxonomy(): + """LIT-4294: a Responses ``input`` list freely mixes message items, + ``function_call`` (no ``role``), and ``function_call_output`` items. The + old ``all(item has 'role')`` gate wrapped the whole list as one blob and + yielded nothing; every text fragment must be visited independently.""" + data = { + "input": [ + { + "type": "message", + "role": "user", + "content": [{"type": "input_text", "text": "hello"}], + }, + { + "type": "function_call", + "call_id": "c1", + "name": "get_weather", + "arguments": "{}", + }, + { + "type": "function_call_output", + "call_id": "c1", + "output": [{"type": "input_text", "text": "sunny"}], + }, + ] + } + assert list(iter_message_text(data)) == ["hello", "sunny"] + + # ── walk_user_text ──────────────────────────────────────────────────────────── @@ -160,6 +208,89 @@ def test_walk_user_text_redacts_responses_api_list_input(): assert data["input"][1] == {"type": "image_url", "image_url": {"url": "..."}} +def test_walk_user_text_redacts_responses_input_text_and_output_text_parts(): + """LIT-4294: ``walk_user_text`` must recognise the Responses text-part + variants so masking guardrails (secret detection, PII) actually redact + ``/v1/responses`` bodies instead of no-op'ing on them.""" + data = { + "input": [ + { + "type": "message", + "role": "user", + "content": [{"type": "input_text", "text": "AKIAEXAMPLE"}], + }, + { + "type": "message", + "role": "assistant", + "content": [{"type": "output_text", "text": "AKIAEXAMPLE too"}], + }, + ] + } + visited = walk_user_text(data, lambda s: s.replace("AKIAEXAMPLE", "[REDACTED]")) + assert visited == 2 + assert data["input"][0]["content"][0] == { + "type": "input_text", + "text": "[REDACTED]", + } + assert data["input"][1]["content"][0] == { + "type": "output_text", + "text": "[REDACTED] too", + } + + +def test_walk_user_text_redacts_function_call_output_text(): + """LIT-4294: tool-call round-trips carry secrets in + ``function_call_output.output``; the redact walker must descend into it + while leaving ``function_call`` items (call_id, arguments) untouched.""" + data = { + "input": [ + { + "type": "message", + "role": "user", + "content": [{"type": "input_text", "text": "AKIAEXAMPLE user"}], + }, + { + "type": "function_call", + "call_id": "c1", + "name": "get_weather", + "arguments": '{"AKIAEXAMPLE": 1}', + }, + { + "type": "function_call_output", + "call_id": "c1", + "output": [{"type": "input_text", "text": "AKIAEXAMPLE tool"}], + }, + ] + } + visited = walk_user_text(data, lambda s: s.replace("AKIAEXAMPLE", "[REDACTED]")) + assert visited == 2 + assert data["input"][0]["content"][0]["text"] == "[REDACTED] user" + assert data["input"][1] == { + "type": "function_call", + "call_id": "c1", + "name": "get_weather", + "arguments": '{"AKIAEXAMPLE": 1}', + } + assert data["input"][2]["output"][0]["text"] == "[REDACTED] tool" + + +def test_walk_user_text_redacts_function_call_output_string_output(): + """LIT-4294: ``function_call_output.output`` is also a plain string in + OpenAI's Responses spec; the redact walker must handle both forms.""" + data = { + "input": [ + { + "type": "function_call_output", + "call_id": "c1", + "output": "AKIAEXAMPLE tool", + }, + ] + } + visited = walk_user_text(data, lambda s: s.replace("AKIAEXAMPLE", "[REDACTED]")) + assert visited == 1 + assert data["input"][0]["output"] == "[REDACTED] tool" + + def test_walk_user_text_redacts_mixed_list_input(): """Read and write helpers must agree on coverage — bare strings inside a mixed ``input`` list are inspected by both.""" @@ -206,17 +337,13 @@ def test_build_inspection_messages_joins_multimodal_text_parts(): } ] } - assert build_inspection_messages(data) == [ - {"role": "user", "content": "first part\nsecond part"} - ] + assert build_inspection_messages(data) == [{"role": "user", "content": "first part\nsecond part"}] def test_build_inspection_messages_lifts_responses_api_input(): """fniVO9-F: ``input`` must be visible to hooks that POST messages to a remote API.""" data = {"input": "responses-api content"} - assert build_inspection_messages(data) == [ - {"role": "user", "content": "responses-api content"} - ] + assert build_inspection_messages(data) == [{"role": "user", "content": "responses-api content"}] def test_build_inspection_messages_drops_messages_with_no_text(): @@ -233,6 +360,102 @@ def test_build_inspection_messages_drops_messages_with_no_text(): assert build_inspection_messages(data) == [{"role": "user", "content": "kept"}] +def test_build_inspection_messages_responses_api_tool_call_taxonomy(): + """LIT-4294: mixed Responses ``input`` (message + function_call + + function_call_output) must produce a non-empty inspection list. The + customer's writeup reproduced a 422 from AIM's ``/fw/v1/analyze`` + (``No messages in the request``) when this synthesised list came back + empty; every other guardrail silently scanned nothing on the same + input.""" + data = { + "input": [ + { + "type": "message", + "role": "user", + "content": [{"type": "input_text", "text": "hello"}], + }, + { + "type": "function_call", + "call_id": "c1", + "name": "get_weather", + "arguments": "{}", + }, + { + "type": "function_call_output", + "call_id": "c1", + "output": [{"type": "input_text", "text": "sunny"}], + }, + ] + } + assert build_inspection_messages(data) == [ + {"role": "user", "content": "hello"}, + {"role": "tool", "content": "sunny"}, + ] + + +def test_build_inspection_messages_function_call_output_defaults_to_tool(): + """LIT-4294: a Responses ``function_call_output`` item is the semantic + equivalent of a chat-completions ``role: "tool"`` message, so the shared + helper synthesises ``role: "tool"`` when the item has no explicit role. + AIM's schema-safe coercion happens at the AIM call site, not here.""" + data = { + "input": [ + { + "type": "function_call_output", + "call_id": "c1", + "output": [{"type": "input_text", "text": "tool text"}], + }, + ] + } + assert build_inspection_messages(data) == [{"role": "tool", "content": "tool text"}] + + +def test_build_inspection_messages_function_call_output_preserves_explicit_role(): + """When ``function_call_output`` carries a caller-supplied ``role`` the + shared helper preserves it rather than synthesising ``tool``.""" + data = { + "input": [ + { + "type": "function_call_output", + "role": "assistant", + "call_id": "c1", + "output": [{"type": "input_text", "text": "tool text"}], + }, + ] + } + assert build_inspection_messages(data) == [{"role": "assistant", "content": "tool text"}] + + +def test_build_inspection_messages_bare_content_part_preserves_explicit_role(): + """A bare content-part dict with an explicit ``role`` keeps it. Only + absent roles get defaulted to ``user``.""" + data = { + "input": [ + {"type": "input_text", "text": "no role"}, + {"type": "output_text", "role": "assistant", "text": "with role"}, + ] + } + assert build_inspection_messages(data) == [ + {"role": "user", "content": "no role"}, + {"role": "assistant", "content": "with role"}, + ] + + +def test_build_inspection_messages_message_item_preserves_role(): + """Responses message items carry a role explicitly; the shared helper + passes it through untouched.""" + data = { + "input": [ + {"type": "message", "role": "system", "content": [{"type": "input_text", "text": "sys"}]}, + {"type": "message", "role": "assistant", "content": [{"type": "output_text", "text": "asst"}]}, + ] + } + assert build_inspection_messages(data) == [ + {"role": "system", "content": "sys"}, + {"role": "assistant", "content": "asst"}, + ] + + def test_build_inspection_messages_empty_data(): assert build_inspection_messages({}) == [] assert build_inspection_messages({"messages": []}) == [] diff --git a/tests/test_litellm/proxy/management_endpoints/test_key_management_endpoints.py b/tests/test_litellm/proxy/management_endpoints/test_key_management_endpoints.py index d707421aeb6..2fe7725fd12 100644 --- a/tests/test_litellm/proxy/management_endpoints/test_key_management_endpoints.py +++ b/tests/test_litellm/proxy/management_endpoints/test_key_management_endpoints.py @@ -9023,7 +9023,7 @@ class TestValidateKeyAliasFormat: litellm.enable_key_alias_format_validation = False def test_validation_skipped_when_flag_disabled(self): - """When enable_key_alias_format_validation is False (default), no validation occurs.""" + """When enable_key_alias_format_validation is False (default), no charset/length validation occurs.""" from litellm.proxy.management_endpoints.key_management_endpoints import ( _validate_key_alias_format, ) @@ -9034,6 +9034,33 @@ class TestValidateKeyAliasFormat: _validate_key_alias_format("!invalid!") _validate_key_alias_format("a" * 256) + @pytest.mark.parametrize( + "unsafe_alias", + [ + "../../../other-app/creds", + "litellm/../../secret", + "foo\n- !grant\n role: !!admin\n member: attacker", + "foo\rbar", + "foo\x00bar", + ], + ) + def test_validate_key_alias_format_rejects_traversal_and_control_chars_even_when_flag_disabled( + self, unsafe_alias + ): + """ + Regression test: this check must reject an invalid key_alias unconditionally, + even when enable_key_alias_format_validation (the separate, opt-in charset + rule) is disabled. + """ + from litellm.proxy.management_endpoints.key_management_endpoints import ( + _validate_key_alias_format, + ) + + with pytest.raises(ProxyException) as exc: + _validate_key_alias_format(unsafe_alias) + assert str(exc.value.code) == "400" + assert "Invalid key_alias" in str(exc.value.message) + def test_validate_key_alias_format_valid(self): from litellm.proxy.management_endpoints.key_management_endpoints import ( _validate_key_alias_format, diff --git a/tests/test_litellm/proxy/management_endpoints/test_mcp_management_endpoints.py b/tests/test_litellm/proxy/management_endpoints/test_mcp_management_endpoints.py index 86bbce36de3..a02acc02502 100644 --- a/tests/test_litellm/proxy/management_endpoints/test_mcp_management_endpoints.py +++ b/tests/test_litellm/proxy/management_endpoints/test_mcp_management_endpoints.py @@ -5134,3 +5134,93 @@ def test_stamp_oauth2_flow_ignores_non_oauth2(): payload = _oauth2_create_payload(auth_type="none") mgmt_endpoints.stamp_omitted_oauth2_flow(payload) assert payload.oauth2_flow is None + + +async def _run_edit(old_record, updated_record, purge_mock=None): + from litellm.proxy.management_endpoints.mcp_management_endpoints import edit_mcp_server + + server_id = updated_record.server_id + with ( + patch("litellm.proxy.management_endpoints.mcp_management_endpoints.MCP_AVAILABLE", True), + patch( + "litellm.proxy.management_endpoints.mcp_management_endpoints.get_prisma_client_or_throw", + return_value=MagicMock(), + ), + patch( + "litellm.proxy.management_endpoints.mcp_management_endpoints.get_mcp_server", + AsyncMock(side_effect=old_record) + if isinstance(old_record, Exception) + else AsyncMock(return_value=old_record), + ), + patch( + "litellm.proxy.management_endpoints.mcp_management_endpoints.update_mcp_server", + AsyncMock(return_value=updated_record), + ), + patch( + "litellm.proxy.management_endpoints.mcp_management_endpoints.validate_and_normalize_mcp_server_payload", + autospec=True, + ), + patch("litellm.proxy.management_endpoints.mcp_management_endpoints.global_mcp_server_manager") as mock_manager, + patch( + "litellm.proxy.management_endpoints.mcp_management_endpoints.purge_user_oauth_credentials_for_server", + purge_mock if purge_mock is not None else AsyncMock(return_value=1), + ) as mock_purge, + ): + mock_manager.update_server = AsyncMock() + mock_manager.reload_servers_from_database = AsyncMock() + payload = UpdateMCPServerRequest(server_id=server_id, alias=updated_record.alias, url=updated_record.url) + user_auth = UserAPIKeyAuth(user_id="admin", user_role=LitellmUserRoles.PROXY_ADMIN) + result = await edit_mcp_server(payload=payload, user_api_key_dict=user_auth) + return result, mock_purge + + +@pytest.mark.asyncio +async def test_edit_mcp_server_purges_user_tokens_on_mint_relevant_change(): + server_id = str(uuid.uuid4()) + old = generate_mock_mcp_server_db_record(server_id=server_id, url="https://old.example.com/mcp") + updated = generate_mock_mcp_server_db_record(server_id=server_id, url="https://new.example.com/mcp") + + result, mock_purge = await _run_edit(old, updated) + + assert result.server_id == server_id + mock_purge.assert_awaited_once() + assert mock_purge.await_args.args[1] == server_id + + +@pytest.mark.asyncio +async def test_edit_mcp_server_skips_purge_when_identity_unchanged(): + server_id = str(uuid.uuid4()) + old = generate_mock_mcp_server_db_record(server_id=server_id, alias="Before") + updated = generate_mock_mcp_server_db_record(server_id=server_id, alias="After") + + result, mock_purge = await _run_edit(old, updated) + + assert result.server_id == server_id + mock_purge.assert_not_awaited() + + +@pytest.mark.asyncio +async def test_edit_mcp_server_purge_failure_does_not_fail_the_edit(): + """The purge is best-effort: a purge exception after a successful update must be swallowed and + logged, never turned into an error response for an edit whose primary job already succeeded.""" + server_id = str(uuid.uuid4()) + old = generate_mock_mcp_server_db_record(server_id=server_id, url="https://old.example.com/mcp") + updated = generate_mock_mcp_server_db_record(server_id=server_id, url="https://new.example.com/mcp") + + result, mock_purge = await _run_edit(old, updated, purge_mock=AsyncMock(side_effect=RuntimeError("db down"))) + + assert result.server_id == server_id + mock_purge.assert_awaited_once() + + +@pytest.mark.asyncio +async def test_edit_mcp_server_snapshot_failure_skips_purge_but_edit_succeeds(): + """The pre-update snapshot read is advisory (it only feeds the purge decision); a read failure + must skip the stale-token check with a warning, never fail the edit itself.""" + server_id = str(uuid.uuid4()) + updated = generate_mock_mcp_server_db_record(server_id=server_id, url="https://new.example.com/mcp") + + result, mock_purge = await _run_edit(RuntimeError("db read failed"), updated) + + assert result.server_id == server_id + mock_purge.assert_not_awaited() diff --git a/tests/test_litellm/proxy/management_endpoints/test_ui_sso.py b/tests/test_litellm/proxy/management_endpoints/test_ui_sso.py index 32229e3e64e..045e15f8b8b 100644 --- a/tests/test_litellm/proxy/management_endpoints/test_ui_sso.py +++ b/tests/test_litellm/proxy/management_endpoints/test_ui_sso.py @@ -389,6 +389,78 @@ async def test_get_user_groups_error_handling(): assert len(result) == 0 +@pytest.mark.asyncio +async def test_get_user_groups_uses_default_graph_endpoint(monkeypatch): + monkeypatch.delenv("MICROSOFT_GRAPH_ENDPOINT", raising=False) + + requested_urls: list[str] = [] + + async def mock_get(url, *args, **kwargs): + requested_urls.append(url) + mock = MagicMock() + mock.json.return_value = {"value": []} + return mock + + with patch( + "litellm.proxy.management_endpoints.ui_sso.get_async_httpx_client" + ) as mock_client: + mock_client.return_value = MagicMock() + mock_client.return_value.get = mock_get + + await MicrosoftSSOHandler.get_user_groups_from_graph_api(access_token="mock_token") + + assert requested_urls == ["https://graph.microsoft.com/v1.0/me/memberOf"] + + +@pytest.mark.asyncio +async def test_get_user_groups_uses_configured_graph_endpoint(monkeypatch): + monkeypatch.setenv("MICROSOFT_GRAPH_ENDPOINT", "https://graph.microsoft.us/v1.0") + + requested_urls: list[str] = [] + + async def mock_get(url, *args, **kwargs): + requested_urls.append(url) + mock = MagicMock() + mock.json.return_value = {"value": []} + return mock + + with patch( + "litellm.proxy.management_endpoints.ui_sso.get_async_httpx_client" + ) as mock_client: + mock_client.return_value = MagicMock() + mock_client.return_value.get = mock_get + + await MicrosoftSSOHandler.get_user_groups_from_graph_api(access_token="mock_token") + + assert requested_urls == ["https://graph.microsoft.us/v1.0/me/memberOf"] + + +@pytest.mark.asyncio +async def test_get_group_ids_from_service_principal_uses_configured_graph_endpoint(monkeypatch): + monkeypatch.setenv("MICROSOFT_GRAPH_ENDPOINT", "https://graph.microsoft.us/v1.0") + + requested_urls: list[str] = [] + + async def mock_get(url, *args, **kwargs): + requested_urls.append(url) + mock = MagicMock() + mock.json.return_value = {"value": []} + return mock + + async_client = MagicMock() + async_client.get = mock_get + + await MicrosoftSSOHandler.get_group_ids_from_service_principal( + service_principal_id="sp-123", + async_client=async_client, + access_token="mock_token", + ) + + assert requested_urls == [ + "https://graph.microsoft.us/v1.0/servicePrincipals/sp-123/appRoleAssignedTo" + ] + + def test_get_group_ids_from_graph_api_response(): # Arrange mock_response = MicrosoftGraphAPIUserGroupResponse( @@ -2711,6 +2783,7 @@ class TestCLIKeyRegenerationFlow: # Mock request mock_request = MagicMock(spec=Request) + mock_request.query_params = {"code": "some-auth-code"} cli_state = f"{LITELLM_CLI_SESSION_TOKEN_PREFIX}:cli-new-session-key-456" @@ -2752,6 +2825,7 @@ class TestCLIKeyRegenerationFlow: from litellm.proxy.management_endpoints.ui_sso import auth_callback mock_request = MagicMock(spec=Request) + mock_request.query_params = {"code": "some-auth-code"} cli_state = ( f"{LITELLM_CLI_SESSION_TOKEN_PREFIX}:cli-new-session-key-456:WXYZ-2345" ) @@ -7182,3 +7256,61 @@ async def test_cli_poll_key_tolerates_missing_user_row(): assert result["status"] == "ready" assert result["key"] == mock_jwt_token assert result["user_id"] == "just-created-user" + + +def _make_sso_callback_request(query_params: dict) -> MagicMock: + mock_request = MagicMock(spec=Request) + mock_request.query_params = query_params + return mock_request + + +@pytest.mark.asyncio +async def test_auth_callback_surfaces_oauth_error_with_description(): + """ + Regression: when the IdP denies access it redirects back with + ?error=...&error_description=... and no `code`. The callback must surface + that reason as a 401 instead of failing later on the missing `code` param. + """ + from litellm.proxy.management_endpoints.ui_sso import auth_callback + + mock_request = _make_sso_callback_request( + {"error": "access_denied", "error_description": "User is not assigned to the client application"} + ) + + with pytest.raises(HTTPException) as exc_info: + await auth_callback(request=mock_request, state=None) + + assert exc_info.value.status_code == 401 + assert "access_denied" in str(exc_info.value.detail) + assert "User is not assigned to the client application" in str(exc_info.value.detail) + + +@pytest.mark.asyncio +async def test_auth_callback_surfaces_oauth_error_without_description(): + """error_description is optional in the OAuth error response; the 401 detail must not render 'None'.""" + from litellm.proxy.management_endpoints.ui_sso import auth_callback + + mock_request = _make_sso_callback_request({"error": "access_denied"}) + + with pytest.raises(HTTPException) as exc_info: + await auth_callback(request=mock_request, state=None) + + assert exc_info.value.status_code == 401 + assert "access_denied" in str(exc_info.value.detail) + assert "None" not in str(exc_info.value.detail) + assert "error_description" not in str(exc_info.value.detail) + + +@pytest.mark.asyncio +async def test_auth_callback_without_oauth_error_proceeds_to_normal_flow(): + """Without an `error` query param the guard must not fire; the callback proceeds into the normal flow.""" + from litellm.proxy.management_endpoints.ui_sso import auth_callback + + mock_request = _make_sso_callback_request({"code": "some-auth-code"}) + + with patch("litellm.proxy.proxy_server.prisma_client", None): + with pytest.raises(HTTPException) as exc_info: + await auth_callback(request=mock_request, state=None) + + assert exc_info.value.status_code == 500 + assert "DB not connected" in str(exc_info.value.detail) diff --git a/tests/test_litellm/proxy/spend_tracking/test_spend_management_endpoints.py b/tests/test_litellm/proxy/spend_tracking/test_spend_management_endpoints.py index 1e9818534c9..c3899883d58 100644 --- a/tests/test_litellm/proxy/spend_tracking/test_spend_management_endpoints.py +++ b/tests/test_litellm/proxy/spend_tracking/test_spend_management_endpoints.py @@ -1,4 +1,5 @@ import asyncio +import collections import datetime import json import os @@ -98,6 +99,7 @@ def _reconstruct_ui_where_from_sql(sql_query, params): alias = re.search(r"user_api_key_alias' LIKE \$(\d+)", cond) code = re.search(r"error_code' = \$(\d+)", cond) msg = re.search(r"error_message' LIKE \$(\d+)", cond) + sess = re.fullmatch(r"session_id LIKE \$(\d+)", cond) status = re.fullmatch(r"status = \$(\d+)", cond) if gte: date_bounds["gte"] = _iso(params[int(gte.group(1)) - 1]) @@ -107,6 +109,8 @@ def _reconstruct_ui_where_from_sql(sql_query, params): where["OR"] = where.get("OR", []) + [{"multi_team": True}] elif "status = 'success'" in cond: where["OR"] = where.get("OR", []) + [{"status": "success"}] + elif sess: + where["session_id"] = {"contains": str(params[int(sess.group(1)) - 1]).strip("%")} elif status: where["status"] = {"equals": params[int(status.group(1)) - 1]} elif alias: @@ -162,7 +166,19 @@ def make_ui_spend_logs_mock_prisma(mock_spend_logs, filter_fn, team_lookup_fn=No async def count(self, *args, **kwargs): return len(filter_fn(kwargs.get("where", {}))) + async def group_by(self, by, where, count): + col = by[0] + allowed = where.get(col, {}).get("in") + tallied = collections.Counter( + log[col] + for log in mock_spend_logs + if log.get(col) is not None and (allowed is None or log[col] in allowed) + ) + return [{col: value, "_count": {col: n}} for value, n in tallied.items()] + async def query_raw(self, sql_query, *params): + if "mcp_tool_call_count" in sql_query: + return [] filtered = filter_fn(_reconstruct_ui_where_from_sql(sql_query, params)) total = len(filtered) if "COUNT(*)" in sql_query: @@ -597,6 +613,73 @@ async def test_ui_view_spend_logs_with_user_id(client, monkeypatch): assert data["data"][0]["user"] == "test_user_1" +@pytest.mark.asyncio +@pytest.mark.parametrize( + "session_id_query,expected_request_ids", + [ + ("session-filter-demo-1", {"req1", "req2"}), + ("session-filter-demo-2", {"req3"}), + ("session-filter", {"req1", "req2", "req3"}), + ("demo", {"req1", "req2", "req3"}), + ("no-such-session", set()), + ], +) +async def test_ui_view_spend_logs_with_session_id( + client, monkeypatch, session_id_query, expected_request_ids +): + def make_log(request_id, session_id): + return { + "id": f"log-{request_id}", + "request_id": request_id, + "api_key": "sk-test-key", + "user": "test_user_1", + "session_id": session_id, + "spend": 0.05, + "startTime": datetime.datetime.now(timezone.utc).isoformat(), + "model": "gpt-4", + } + + mock_spend_logs = [ + make_log("req1", "session-filter-demo-1"), + make_log("req2", "session-filter-demo-1"), + make_log("req3", "session-filter-demo-2"), + make_log("req4", "unrelated-abc"), + ] + + def filter_by_session(where): + session_filter = where.get("session_id") + if session_filter is None: + return mock_spend_logs + return [ + log + for log in mock_spend_logs + if session_filter["contains"] in log["session_id"] + ] + + monkeypatch.setattr( + "litellm.proxy.proxy_server.prisma_client", + make_ui_spend_logs_mock_prisma(mock_spend_logs, filter_by_session), + ) + + start_date, end_date = _default_date_range() + + response = client.get( + "/spend/logs/ui", + params={ + "session_id": session_id_query, + "start_date": start_date, + "end_date": end_date, + }, + headers={"Authorization": "Bearer sk-test"}, + ) + + assert response.status_code == 200 + data = response.json() + assert data["total"] == len(expected_request_ids) + assert {log["request_id"] for log in data["data"]} == expected_request_ids + assert all(session_id_query in log["session_id"] for log in data["data"]) + + # Mock spend logs with distinct values for sorting tests. # req_a: spend=0.10, tokens=500, start/end earliest # req_b: spend=0.05, tokens=200, start/end 2nd diff --git a/tests/test_litellm/rerank_api/__init__.py b/tests/test_litellm/rerank_api/__init__.py new file mode 100644 index 00000000000..e69de29bb2d diff --git a/tests/test_litellm/rerank_api/test_main.py b/tests/test_litellm/rerank_api/test_main.py new file mode 100644 index 00000000000..46d1461da50 --- /dev/null +++ b/tests/test_litellm/rerank_api/test_main.py @@ -0,0 +1,67 @@ +import logging +import os +import sys +from unittest.mock import MagicMock, patch + +sys.path.insert(0, os.path.abspath("../../..")) + +import litellm + +MARKER_QUERY = "MARKER_QUERY_do_not_log_at_info" +MARKER_DOC = "MARKER_DOC_sensitive_customer_text" + + +def _mock_cohere_response() -> MagicMock: + mock_response = MagicMock() + + def return_val(): + return { + "id": "cmpl-mockid", + "results": [{"index": 0, "relevance_score": 0.95}], + "meta": { + "api_version": {"version": "1.0"}, + "billed_units": {"search_units": 1}, + }, + } + + mock_response.json = return_val + mock_response.headers = {"key": "value"} + mock_response.status_code = 200 + return mock_response + + +def test_rerank_does_not_log_request_content_at_info(caplog): + """Regression for #32525: rerank must not emit query/documents to logs at INFO. + + The mapped ``optional_rerank_params`` (which always contains ``query`` and + ``documents``) bypasses ``turn_off_message_logging`` / ``redact_messages``, + so logging it at INFO leaks raw request content into stdout and any log sink. + """ + litellm.cohere_key = "test_api_key" + caplog.set_level(logging.DEBUG, logger="LiteLLM") + + with patch( + "litellm.llms.custom_httpx.http_handler.HTTPHandler.post", + return_value=_mock_cohere_response(), + ): + litellm.rerank( + model="cohere/rerank-english-v3.0", + query=MARKER_QUERY, + documents=[MARKER_DOC, "unrelated"], + top_n=2, + ) + + litellm_records = [r for r in caplog.records if r.name == "LiteLLM"] + + info_or_above = [ + r.getMessage() + for r in litellm_records + if r.levelno >= logging.INFO and (MARKER_QUERY in r.getMessage() or MARKER_DOC in r.getMessage()) + ] + assert not info_or_above, f"rerank leaked request content at INFO+: {info_or_above}" + + optional_params_logs = [r for r in litellm_records if "optional_rerank_params" in r.getMessage()] + assert optional_params_logs, "expected the optional_rerank_params line to be logged" + assert all( + r.levelno == logging.DEBUG for r in optional_params_logs + ), "optional_rerank_params must be logged at DEBUG, not INFO" diff --git a/tests/test_litellm/responses/test_responses_api_request_body.py b/tests/test_litellm/responses/test_responses_api_request_body.py index e312a11e893..c39ba75bd97 100644 --- a/tests/test_litellm/responses/test_responses_api_request_body.py +++ b/tests/test_litellm/responses/test_responses_api_request_body.py @@ -1,6 +1,7 @@ """ Test that litellm.responses() / litellm.aresponses() send the expected request body -over the wire. Expected JSON bodies are stored in expected_responses_api_request/. +over the wire and surface provider errors correctly. Expected JSON bodies are stored +in expected_responses_api_request/. """ import json @@ -18,24 +19,20 @@ def _expected_dir() -> Path: return Path(__file__).resolve().parent.parent / "expected_responses_api_request" -@pytest.mark.asyncio -async def test_aresponses_context_management_and_shell_request_body_matches_expected(): - """ - Call litellm.aresponses() with context_management and shell tool; - assert the httpx POST request body matches the expected JSON. - """ - expected_path = _expected_dir() / "context_management_and_shell.json" +def _load_expected_body(filename: str) -> dict: + expected_path = _expected_dir() / filename assert expected_path.exists(), f"Expected file not found: {expected_path}" with open(expected_path) as f: - expected_body = json.load(f) + return json.load(f) - # Minimal Responses API response so parsing succeeds - mock_response = { - "id": "resp_ctx_shell_test", + +def _minimal_responses_api_payload(response_id: str, model: str) -> dict: + return { + "id": response_id, "object": "response", "created_at": 1734366691, "status": "completed", - "model": "gpt-4o", + "model": model, "output": [ { "type": "message", @@ -69,21 +66,41 @@ async def test_aresponses_context_management_and_shell_request_body_matches_expe "user": None, } - class MockResponse: - def __init__(self, json_data, status_code=200): - self._json_data = json_data - self.status_code = status_code - self.text = json.dumps(json_data) - self.headers = httpx.Headers({}) - def json(self): - return self._json_data +class MockResponse: + def __init__(self, json_data, status_code=200): + self._json_data = json_data + self.status_code = status_code + self.text = json.dumps(json_data) + self.headers = httpx.Headers({}) + + def json(self): + return self._json_data + + +def _assert_request_body_matches(request_body: dict, expected_body: dict) -> None: + for key, expected_value in expected_body.items(): + assert key in request_body, f"Missing key in request body: {key}" + assert ( + request_body[key] == expected_value + ), f"Mismatch for key {key}: got {request_body[key]!r}, expected {expected_value!r}" + + +@pytest.mark.asyncio +async def test_aresponses_context_management_and_shell_request_body_matches_expected(): + """ + Call litellm.aresponses() with context_management and shell tool; + assert the httpx POST request body matches the expected JSON. + """ + expected_body = _load_expected_body("context_management_and_shell.json") with patch( "litellm.llms.custom_httpx.http_handler.AsyncHTTPHandler.post", new_callable=AsyncMock, ) as mock_post: - mock_post.return_value = MockResponse(mock_response, 200) + mock_post.return_value = MockResponse( + _minimal_responses_api_payload("resp_ctx_shell_test", "gpt-4o"), 200 + ) await litellm.aresponses( model="openai/gpt-4o", @@ -95,10 +112,87 @@ async def test_aresponses_context_management_and_shell_request_body_matches_expe ) mock_post.assert_called_once() - request_body = mock_post.call_args.kwargs["json"] + _assert_request_body_matches(mock_post.call_args.kwargs["json"], expected_body) - for key, expected_value in expected_body.items(): - assert key in request_body, f"Missing key in request body: {key}" - assert ( - request_body[key] == expected_value - ), f"Mismatch for key {key}: got {request_body[key]!r}, expected {expected_value!r}" + +@pytest.mark.asyncio +async def test_aresponses_azure_shell_tool_request_body_matches_expected(): + """ + Call litellm.aresponses() on the Azure route with the shell tool; + assert the httpx POST request body carries the shell tool verbatim. + """ + expected_body = _load_expected_body("azure_shell_tool.json") + + with patch( + "litellm.llms.custom_httpx.http_handler.AsyncHTTPHandler.post", + new_callable=AsyncMock, + ) as mock_post: + mock_post.return_value = MockResponse( + _minimal_responses_api_payload("resp_azure_shell_test", "gpt-5-mini"), 200 + ) + + await litellm.aresponses( + model="azure/gpt-5-mini", + api_base="https://fake-resource.openai.azure.com", + api_key="fake-api-key", + api_version="2025-03-01-preview", + input=expected_body["input"], + tools=expected_body["tools"], + tool_choice=expected_body["tool_choice"], + max_output_tokens=expected_body["max_output_tokens"], + ) + + mock_post.assert_called_once() + _assert_request_body_matches(mock_post.call_args.kwargs["json"], expected_body) + + +@pytest.mark.asyncio +async def test_aresponses_azure_shell_tool_400_maps_to_bad_request_error(): + """ + Azure rejects the shell tool for unsupported deployments with a 400; + litellm must surface that as litellm.BadRequestError carrying the provider message. + """ + error_body = { + "error": { + "message": "Tool of type 'shell' is not supported with this model.", + "type": "invalid_request_error", + "param": "tools", + "code": None, + } + } + + def _raise_azure_400(*args, **kwargs): + response = httpx.Response( + status_code=400, + json=error_body, + request=httpx.Request( + "POST", + kwargs.get( + "url", + "https://fake-resource.openai.azure.com/openai/responses", + ), + ), + ) + response.raise_for_status() + + with patch( + "litellm.llms.custom_httpx.http_handler.AsyncHTTPHandler.post", + new_callable=AsyncMock, + ) as mock_post: + mock_post.side_effect = _raise_azure_400 + + with pytest.raises(litellm.BadRequestError) as excinfo: + await litellm.aresponses( + model="azure/gpt-5-mini", + api_base="https://fake-resource.openai.azure.com", + api_key="fake-api-key", + api_version="2025-03-01-preview", + input="List files in /mnt/data and run python --version.", + tools=[{"type": "shell", "environment": {"type": "container_auto"}}], + tool_choice="auto", + max_output_tokens=256, + ) + + assert excinfo.value.status_code == 400 + assert "shell" in str(excinfo.value).lower() + assert "not supported" in str(excinfo.value).lower() diff --git a/tests/test_litellm/secret_managers/test_base_secret_manager.py b/tests/test_litellm/secret_managers/test_base_secret_manager.py new file mode 100644 index 00000000000..cba6a99ab7f --- /dev/null +++ b/tests/test_litellm/secret_managers/test_base_secret_manager.py @@ -0,0 +1,59 @@ +""" +Test raise_if_unsafe_secret_name, the shared guard applied before secret_name +reaches a secret manager backend. +""" + +import os +import sys + +import pytest + +sys.path.insert(0, os.path.abspath("../..")) # Adds the parent directory to the system path + +from litellm.secret_managers.base_secret_manager import raise_if_unsafe_secret_name + + +@pytest.mark.parametrize( + "secret_name", + [ + "..", + "../../../other-app/creds", + "litellm/../../secret", + "foo/../bar", + "foo/..", + "../foo", + "foo\nbar", + "foo\rbar", + "foo\x00bar", + "foo\x7fbar", + "foo\x85bar", + "foo
bar", + "foo
bar", + ], +) +def test_raise_if_unsafe_secret_name_rejects_traversal_and_line_breaks(secret_name): + with pytest.raises(ValueError): + raise_if_unsafe_secret_name(secret_name) + + +@pytest.mark.parametrize( + "secret_name", + [ + "plain-alias", + "my-key-123", + "prod/my-service-key", + "team/user@example.com", + "foo: bar", + "foo # bar", + "foo?evil=1", + "foo#bar", + "a" * 500, + "release-1.0..2", + "my..key", + "..foo", + "foo..", + "v2.0..1-beta", + ], +) +def test_raise_if_unsafe_secret_name_allows_legitimate_aliases(secret_name): + raise_if_unsafe_secret_name(secret_name) diff --git a/tests/test_litellm/test_gpt_5_6_model_metadata.py b/tests/test_litellm/test_gpt_5_6_model_metadata.py new file mode 100644 index 00000000000..af9bb117f78 --- /dev/null +++ b/tests/test_litellm/test_gpt_5_6_model_metadata.py @@ -0,0 +1,150 @@ +import json +from pathlib import Path + +import pytest + +from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider + +GPT_5_6_MODELS = ("gpt-5.6", "gpt-5.6-sol", "gpt-5.6-terra", "gpt-5.6-luna") + +STANDARD_PRICING = { + "gpt-5.6": (5e-06, 3e-05, 5e-07, 6.25e-06), + "gpt-5.6-sol": (5e-06, 3e-05, 5e-07, 6.25e-06), + "gpt-5.6-terra": (2.5e-06, 1.5e-05, 2.5e-07, 3.125e-06), + "gpt-5.6-luna": (1e-06, 6e-06, 1e-07, 1.25e-06), +} + + +@pytest.mark.parametrize("model", GPT_5_6_MODELS) +def test_openai_gpt_5_6_model_info(model): + json_path = Path(__file__).parents[2] / "model_prices_and_context_window.json" + with open(json_path) as f: + model_cost = json.load(f) + + info = model_cost.get(model) + assert info is not None, f"{model} not found in model_prices_and_context_window.json" + + assert info["litellm_provider"] == "openai" + assert info["mode"] == "chat" + + input_cost, output_cost, cache_read_cost, cache_write_cost = STANDARD_PRICING[model] + assert info["input_cost_per_token"] == input_cost + assert info["output_cost_per_token"] == output_cost + assert info["cache_read_input_token_cost"] == cache_read_cost + assert info["cache_creation_input_token_cost"] == cache_write_cost + assert info["cache_creation_input_token_cost"] == pytest.approx(input_cost * 1.25) + + assert info["input_cost_per_token_above_272k_tokens"] == pytest.approx(input_cost * 2) + assert info["output_cost_per_token_above_272k_tokens"] == pytest.approx(output_cost * 1.5) + assert info["cache_read_input_token_cost_above_272k_tokens"] == pytest.approx(cache_read_cost * 2) + + assert info["max_input_tokens"] == 1050000 + assert info["max_output_tokens"] == 128000 + assert info["max_tokens"] == 128000 + + assert info["supports_function_calling"] is True + assert info["supports_prompt_caching"] is True + assert info["supports_reasoning"] is True + assert info["supports_response_schema"] is True + assert info["supports_tool_choice"] is True + assert info["supports_vision"] is True + assert info["supports_web_search"] is True + assert info["supports_none_reasoning_effort"] is True + assert info["supports_xhigh_reasoning_effort"] is True + assert info["supports_minimal_reasoning_effort"] is False + + assert info["supported_endpoints"] == ["/v1/chat/completions", "/v1/batch", "/v1/responses"] + assert info["supported_modalities"] == ["text", "image"] + assert info["supported_output_modalities"] == ["text"] + + routed_model, provider, _, _ = get_llm_provider(model=f"openai/{model}") + assert routed_model == model + assert provider == "openai" + + +AZURE_GLOBAL_MODELS = ( + "azure/gpt-5.6", + "azure/gpt-5.6-sol", + "azure/gpt-5.6-terra", + "azure/gpt-5.6-luna", +) + +AZURE_REGIONAL_MODELS = tuple( + f"azure/{region}/{tier}" + for region in ("us", "eu") + for tier in ("gpt-5.6", "gpt-5.6-sol", "gpt-5.6-terra", "gpt-5.6-luna") +) + + +def _tier_key(azure_model): + return azure_model.split("/")[-1] + + +def _load_main(): + json_path = Path(__file__).parents[2] / "model_prices_and_context_window.json" + with open(json_path) as f: + return json.load(f) + + +@pytest.mark.parametrize("model", AZURE_GLOBAL_MODELS) +def test_azure_gpt_5_6_global_model_info(model): + model_cost = _load_main() + info = model_cost.get(model) + assert info is not None, f"{model} not found in model_prices_and_context_window.json" + + assert info["litellm_provider"] == "azure" + assert info["mode"] == "chat" + + input_cost, output_cost, cache_read_cost, _ = STANDARD_PRICING[_tier_key(model)] + assert info["input_cost_per_token"] == input_cost + assert info["output_cost_per_token"] == output_cost + assert info["cache_read_input_token_cost"] == cache_read_cost + + assert info["input_cost_per_token_above_272k_tokens"] == pytest.approx(input_cost * 2) + assert info["output_cost_per_token_above_272k_tokens"] == pytest.approx(output_cost * 1.5) + assert info["input_cost_per_token_priority"] == pytest.approx(input_cost * 2) + assert info["output_cost_per_token_priority"] == pytest.approx(output_cost * 2) + assert info["input_cost_per_token_above_272k_tokens_priority"] == pytest.approx(input_cost * 4) + assert info["output_cost_per_token_above_272k_tokens_priority"] == pytest.approx(output_cost * 3) + + assert info["max_input_tokens"] == 1050000 + assert info["max_output_tokens"] == 128000 + assert info["supports_reasoning"] is True + + routed_model, provider, _, _ = get_llm_provider(model=model) + assert provider == "azure" + + +@pytest.mark.parametrize("model", AZURE_REGIONAL_MODELS) +def test_azure_gpt_5_6_regional_model_info(model): + model_cost = _load_main() + info = model_cost.get(model) + assert info is not None, f"{model} not found in model_prices_and_context_window.json" + + assert info["litellm_provider"] == "azure" + input_cost, output_cost, cache_read_cost, _ = STANDARD_PRICING[_tier_key(model)] + + assert info["input_cost_per_token"] == pytest.approx(input_cost * 1.1) + assert info["output_cost_per_token"] == pytest.approx(output_cost * 1.1) + assert info["cache_read_input_token_cost"] == pytest.approx(cache_read_cost * 1.1) + assert info["input_cost_per_token_above_272k_tokens"] == pytest.approx(input_cost * 2.2) + assert info["output_cost_per_token_above_272k_tokens"] == pytest.approx(output_cost * 1.65) + assert info["input_cost_per_token_priority"] == pytest.approx(input_cost * 2.75) + assert info["output_cost_per_token_priority"] == pytest.approx(output_cost * 2.75) + + +def test_gpt_5_6_backup_matches_main(): + """Ensure the bundled model cost map stays in sync with the canonical file.""" + repo_root = Path(__file__).parents[2] + main_path = repo_root / "model_prices_and_context_window.json" + backup_path = repo_root / "litellm" / "model_prices_and_context_window_backup.json" + + with open(main_path) as f: + main_cost = json.load(f) + with open(backup_path) as f: + backup_cost = json.load(f) + + for model in GPT_5_6_MODELS + AZURE_GLOBAL_MODELS + AZURE_REGIONAL_MODELS: + assert backup_cost.get(model) == main_cost.get(model), ( + f"{model} differs between main and backup model cost maps" + ) diff --git a/tests/test_litellm/test_muse_spark_1_1_model_metadata.py b/tests/test_litellm/test_muse_spark_1_1_model_metadata.py new file mode 100644 index 00000000000..540b97884dc --- /dev/null +++ b/tests/test_litellm/test_muse_spark_1_1_model_metadata.py @@ -0,0 +1,63 @@ +import json +from pathlib import Path + +from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider + +MUSE_SPARK_MODEL = "meta/muse-spark-1.1" + + +def test_muse_spark_1_1_model_info(): + json_path = Path(__file__).parents[2] / "model_prices_and_context_window.json" + with open(json_path) as f: + model_cost = json.load(f) + + info = model_cost.get(MUSE_SPARK_MODEL) + assert info is not None, f"{MUSE_SPARK_MODEL} not found in model_prices_and_context_window.json" + + assert info["litellm_provider"] == "meta" + assert info["mode"] == "chat" + + assert info["input_cost_per_token"] == 1.25e-06 + assert info["output_cost_per_token"] == 4.25e-06 + assert info["cache_read_input_token_cost"] == 1.5e-07 + + assert info["max_input_tokens"] == 1048576 + assert info["max_output_tokens"] == 131072 + assert info["max_tokens"] == 131072 + + assert info["supports_function_calling"] is True + assert info["supports_parallel_function_calling"] is True + assert info["supports_prompt_caching"] is True + assert info["supports_reasoning"] is True + assert info["supports_response_schema"] is True + assert info["supports_tool_choice"] is True + assert info["supports_vision"] is True + assert info["supports_pdf_input"] is True + assert info["supports_web_search"] is True + assert info["supports_minimal_reasoning_effort"] is True + assert info["supports_xhigh_reasoning_effort"] is True + + assert info["supported_endpoints"] == ["/v1/chat/completions", "/v1/responses", "/v1/messages"] + assert info["supported_modalities"] == ["text", "image", "video"] + assert info["supported_output_modalities"] == ["text"] + + routed_model, provider, _, api_base = get_llm_provider(model=MUSE_SPARK_MODEL, api_key="sk-test") + assert routed_model == "muse-spark-1.1" + assert provider == "meta" + assert api_base == "https://api.meta.ai/v1" + + +def test_muse_spark_1_1_backup_matches_main(): + """Ensure the bundled model cost map stays in sync with the canonical file.""" + repo_root = Path(__file__).parents[2] + main_path = repo_root / "model_prices_and_context_window.json" + backup_path = repo_root / "litellm" / "model_prices_and_context_window_backup.json" + + with open(main_path) as f: + main_cost = json.load(f) + with open(backup_path) as f: + backup_cost = json.load(f) + + assert backup_cost.get(MUSE_SPARK_MODEL) == main_cost.get(MUSE_SPARK_MODEL), ( + f"{MUSE_SPARK_MODEL} differs between main and backup model cost maps" + ) diff --git a/tests/test_litellm/test_utils.py b/tests/test_litellm/test_utils.py index 053fe970d3e..d739f9c116a 100644 --- a/tests/test_litellm/test_utils.py +++ b/tests/test_litellm/test_utils.py @@ -710,6 +710,9 @@ def test_aaamodel_prices_and_context_window_json_is_valid(): "cache_creation_input_token_cost": {"type": "number"}, "cache_creation_input_token_cost_above_1hr": {"type": "number"}, "cache_creation_input_token_cost_above_200k_tokens": {"type": "number"}, + "cache_creation_input_token_cost_above_272k_tokens": {"type": "number"}, + "cache_creation_input_token_cost_flex": {"type": "number"}, + "cache_creation_input_token_cost_priority": {"type": "number"}, "cache_read_input_token_cost": {"type": "number"}, "cache_read_input_token_cost_above_200k_tokens": {"type": "number"}, "cache_read_input_token_cost_above_272k_tokens": {"type": "number"}, @@ -871,6 +874,7 @@ def test_aaamodel_prices_and_context_window_json_is_valid(): "/v1/embeddings", "/v1/chat/completions", "/v1/completions", + "/v1/messages", "/v1/images/generations", "/v1/realtime", "/v1/realtime/transcription_sessions", @@ -881,7 +885,6 @@ def test_aaamodel_prices_and_context_window_json_is_valid(): "/v1/audio/speech", "/v1/ocr", "/vertex_ai/live", - "/v1/realtime/transcription_sessions", ], }, }, diff --git a/ui/litellm-dashboard/eslint-budgets.json b/ui/litellm-dashboard/eslint-budgets.json index 8dedb9ac9ca..f08e1bb6160 100644 --- a/ui/litellm-dashboard/eslint-budgets.json +++ b/ui/litellm-dashboard/eslint-budgets.json @@ -2,5 +2,7 @@ "@typescript-eslint/no-explicit-any": { "max": 2040, "target": 1500 }, "no-console": { "max": 484, "target": 0 }, "complexity": { "max": 140, "target": 80 }, - "max-depth": { "max": 70, "target": 30 } + "max-depth": { "max": 70, "target": 30 }, + "local/no-large-inline-object-arg": { "max": 560, "target": 300 }, + "local/no-long-condition-chain": { "max": 265, "target": 120 } } diff --git a/ui/litellm-dashboard/eslint-metrics.json b/ui/litellm-dashboard/eslint-metrics.json index 219cb0580e7..2e204c63a48 100644 --- a/ui/litellm-dashboard/eslint-metrics.json +++ b/ui/litellm-dashboard/eslint-metrics.json @@ -1,6 +1,8 @@ { - "@typescript-eslint/no-explicit-any": 1990, - "complexity": 128, + "@typescript-eslint/no-explicit-any": 1978, + "complexity": 129, + "local/no-large-inline-object-arg": 512, + "local/no-long-condition-chain": 233, "max-depth": 59, "no-console": 15 } diff --git a/ui/litellm-dashboard/eslint-suppressions.json b/ui/litellm-dashboard/eslint-suppressions.json index 1c8f92b720f..200e1175c7f 100644 --- a/ui/litellm-dashboard/eslint-suppressions.json +++ b/ui/litellm-dashboard/eslint-suppressions.json @@ -1,30 +1,43 @@ { - "src/app/(dashboard)/api-reference/APIReferenceView.tsx": { + "scripts/check-lint-budgets.mjs": { + "no-nested-ternary": { + "count": 1 + } + }, + "src/app/(dashboard)/admin-panel/_components/AdminPanel.tsx": { + "no-restricted-imports": { + "count": 1 + }, + "react-hooks/set-state-in-effect": { + "count": 1 + } + }, + "src/app/(dashboard)/api-reference/_components/APIReferenceView.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/budgets/components/budget_modal.tsx": { + "src/app/(dashboard)/budgets/_components/budget_modal.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/budgets/components/budget_panel.test.tsx": { + "src/app/(dashboard)/budgets/_components/budget_panel.test.tsx": { "unused-imports/no-unused-imports": { "count": 2 } }, - "src/app/(dashboard)/budgets/components/budget_panel.tsx": { + "src/app/(dashboard)/budgets/_components/budget_panel.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/budgets/components/edit_budget_modal.tsx": { + "src/app/(dashboard)/budgets/_components/edit_budget_modal.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/caching/components/cache_dashboard.tsx": { + "src/app/(dashboard)/caching/_components/cache_dashboard.tsx": { "no-restricted-imports": { "count": 1 }, @@ -35,17 +48,17 @@ "count": 2 } }, - "src/app/(dashboard)/caching/components/cache_health.tsx": { + "src/app/(dashboard)/caching/_components/cache_health.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/caching/components/cache_settings/RedisTypeSelector.tsx": { + "src/app/(dashboard)/caching/_components/cache_settings/RedisTypeSelector.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/caching/components/cache_settings/index.tsx": { + "src/app/(dashboard)/caching/_components/cache_settings/index.tsx": { "no-restricted-imports": { "count": 1 }, @@ -53,99 +66,102 @@ "count": 1 } }, - "src/app/(dashboard)/cost-tracking/components/add_margin_form.tsx": { + "src/app/(dashboard)/cost-tracking/_components/add_margin_form.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/cost-tracking/components/add_provider_form.tsx": { + "src/app/(dashboard)/cost-tracking/_components/add_provider_form.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/cost-tracking/components/cost_tracking_settings.tsx": { + "src/app/(dashboard)/cost-tracking/_components/cost_tracking_settings.tsx": { + "no-nested-ternary": { + "count": 2 + }, "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/cost-tracking/components/how_it_works.tsx": { + "src/app/(dashboard)/cost-tracking/_components/how_it_works.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_cost_results.test.tsx": { + "src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_cost_results.test.tsx": { "unused-imports/no-unused-imports": { "count": 1 } }, - "src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_cost_results.tsx": { + "src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_cost_results.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_export_dropdown.test.tsx": { + "src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_export_dropdown.test.tsx": { "unused-imports/no-unused-imports": { "count": 1 } }, - "src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_export_dropdown.tsx": { + "src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_export_dropdown.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/cost-tracking/components/pricing_calculator/use_multi_cost_estimate.ts": { + "src/app/(dashboard)/cost-tracking/_components/pricing_calculator/use_multi_cost_estimate.ts": { "no-restricted-syntax": { "count": 1 } }, - "src/app/(dashboard)/cost-tracking/components/provider_discount_table.test.tsx": { + "src/app/(dashboard)/cost-tracking/_components/provider_discount_table.test.tsx": { "unused-imports/no-unused-imports": { "count": 1 } }, - "src/app/(dashboard)/cost-tracking/components/provider_discount_table.tsx": { + "src/app/(dashboard)/cost-tracking/_components/provider_discount_table.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/cost-tracking/components/provider_display_helpers.test.ts": { + "src/app/(dashboard)/cost-tracking/_components/provider_display_helpers.test.ts": { "unused-imports/no-unused-imports": { "count": 1 } }, - "src/app/(dashboard)/cost-tracking/components/provider_margin_table.tsx": { + "src/app/(dashboard)/cost-tracking/_components/provider_margin_table.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/cost-tracking/components/use_discount_config.ts": { + "src/app/(dashboard)/cost-tracking/_components/use_discount_config.ts": { "no-restricted-syntax": { "count": 2 } }, - "src/app/(dashboard)/cost-tracking/components/use_margin_config.ts": { + "src/app/(dashboard)/cost-tracking/_components/use_margin_config.ts": { "no-restricted-syntax": { "count": 2 } }, - "src/app/(dashboard)/guardrails-monitor/components/EvaluationSettingsModal.tsx": { + "src/app/(dashboard)/guardrails-monitor/_components/EvaluationSettingsModal.tsx": { "react-hooks/set-state-in-effect": { "count": 1 } }, - "src/app/(dashboard)/guardrails-monitor/components/GuardrailsMonitorView.tsx": { + "src/app/(dashboard)/guardrails-monitor/_components/GuardrailDetail.tsx": { + "no-nested-ternary": { + "count": 3 + } + }, + "src/app/(dashboard)/guardrails-monitor/_components/GuardrailsMonitorView.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/guardrails-monitor/components/ScoreChart.test.tsx": { - "react/display-name": { - "count": 1 - } - }, - "src/app/(dashboard)/guardrails-monitor/components/ScoreChart.tsx": { - "no-restricted-imports": { - "count": 1 + "src/app/(dashboard)/guardrails-monitor/_components/GuardrailsOverview.tsx": { + "no-nested-ternary": { + "count": 8 } }, "src/app/(dashboard)/hooks/accessGroups/useAccessGroupDetails.ts": { @@ -308,7 +324,7 @@ "count": 2 } }, - "src/app/(dashboard)/memory/components/MemoryView.tsx": { + "src/app/(dashboard)/memory/_components/MemoryView.tsx": { "react-hooks/set-state-in-effect": { "count": 1 } @@ -343,6 +359,9 @@ } }, "src/app/(dashboard)/models-and-endpoints/components/ModelRetrySettingsTab.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } @@ -352,6 +371,22 @@ "count": 1 } }, + "src/app/(dashboard)/old-usage/_components/usage.tsx": { + "no-restricted-imports": { + "count": 2 + }, + "react-hooks/immutability": { + "count": 1 + }, + "react-hooks/purity": { + "count": 1 + } + }, + "src/app/(dashboard)/organizations/_components/organizations.tsx": { + "no-restricted-imports": { + "count": 1 + } + }, "src/app/(dashboard)/playground/components/chat_ui/AdditionalModelSettings.tsx": { "no-restricted-imports": { "count": 1 @@ -361,6 +396,9 @@ } }, "src/app/(dashboard)/playground/components/chat_ui/AgentBuilderView.tsx": { + "no-nested-ternary": { + "count": 2 + }, "react-hooks/set-state-in-effect": { "count": 5 } @@ -370,7 +408,15 @@ "count": 1 } }, + "src/app/(dashboard)/playground/components/chat_ui/ChatMessageBubble.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, "src/app/(dashboard)/playground/components/chat_ui/ChatUI.tsx": { + "no-nested-ternary": { + "count": 7 + }, "no-restricted-imports": { "count": 1 }, @@ -382,6 +428,9 @@ } }, "src/app/(dashboard)/playground/components/chat_ui/CodeInterpreterOutput.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-syntax": { "count": 2 } @@ -392,6 +441,9 @@ } }, "src/app/(dashboard)/playground/components/chat_ui/RealtimePlayground.tsx": { + "no-nested-ternary": { + "count": 2 + }, "react-hooks/immutability": { "count": 2 }, @@ -400,16 +452,27 @@ } }, "src/app/(dashboard)/playground/components/compareUI/CompareUI.tsx": { + "no-nested-ternary": { + "count": 4 + }, "react-hooks/set-state-in-effect": { "count": 1 } }, + "src/app/(dashboard)/playground/components/compareUI/components/MessageDisplay.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, "src/app/(dashboard)/playground/components/compareUI/components/ModelSelector.tsx": { "no-restricted-imports": { "count": 1 } }, "src/app/(dashboard)/playground/components/complianceUI/ComplianceUI.tsx": { + "no-nested-ternary": { + "count": 8 + }, "react-hooks/preserve-manual-memoization": { "count": 3 } @@ -473,32 +536,38 @@ "count": 1 } }, - "src/app/(dashboard)/projects/components/ProjectDetailsPage.tsx": { + "src/app/(dashboard)/projects/_components/ProjectDetailsPage.tsx": { + "no-nested-ternary": { + "count": 3 + }, "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/projects/components/ProjectKeysSection.tsx": { + "src/app/(dashboard)/projects/_components/ProjectKeysSection.tsx": { "react-hooks/set-state-in-effect": { "count": 1 } }, - "src/app/(dashboard)/projects/components/ProjectModals/ProjectBaseForm.tsx": { + "src/app/(dashboard)/projects/_components/ProjectModals/ProjectBaseForm.tsx": { "react-hooks/set-state-in-effect": { "count": 2 } }, - "src/app/(dashboard)/projects/components/ProjectsPage.tsx": { + "src/app/(dashboard)/projects/_components/ProjectsPage.tsx": { "react-hooks/set-state-in-effect": { "count": 1 } }, - "src/app/(dashboard)/prompts/components/add_prompt_form.tsx": { + "src/app/(dashboard)/prompts/_components/add_prompt_form.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/prompts/components/index.tsx": { + "src/app/(dashboard)/prompts/_components/index.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 }, @@ -506,17 +575,20 @@ "count": 1 } }, - "src/app/(dashboard)/prompts/components/prompt_editor_view/DeveloperMessageCard.tsx": { + "src/app/(dashboard)/prompts/_components/prompt_editor_view/DeveloperMessageCard.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/prompts/components/prompt_editor_view/ModelConfigCard.tsx": { + "src/app/(dashboard)/prompts/_components/prompt_editor_view/ModelConfigCard.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/prompts/components/prompt_editor_view/PromptCodeSnippets.tsx": { + "src/app/(dashboard)/prompts/_components/prompt_editor_view/PromptCodeSnippets.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 }, @@ -524,52 +596,58 @@ "count": 1 } }, - "src/app/(dashboard)/prompts/components/prompt_editor_view/PromptEditorHeader.tsx": { + "src/app/(dashboard)/prompts/_components/prompt_editor_view/PromptEditorHeader.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/prompts/components/prompt_editor_view/PromptMessagesCard.tsx": { + "src/app/(dashboard)/prompts/_components/prompt_editor_view/PromptMessagesCard.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/prompts/components/prompt_editor_view/PublishModal.tsx": { + "src/app/(dashboard)/prompts/_components/prompt_editor_view/PublishModal.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/prompts/components/prompt_editor_view/ToolsCard.tsx": { + "src/app/(dashboard)/prompts/_components/prompt_editor_view/ToolsCard.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/prompts/components/prompt_editor_view/VersionHistorySidePanel.test.tsx": { + "src/app/(dashboard)/prompts/_components/prompt_editor_view/VersionHistorySidePanel.test.tsx": { "max-nested-callbacks": { "count": 1 } }, - "src/app/(dashboard)/prompts/components/prompt_editor_view/VersionHistorySidePanel.tsx": { + "src/app/(dashboard)/prompts/_components/prompt_editor_view/VersionHistorySidePanel.tsx": { + "no-nested-ternary": { + "count": 1 + }, "react-hooks/immutability": { "count": 1 } }, - "src/app/(dashboard)/prompts/components/prompt_editor_view/conversation_panel/MessageInput.tsx": { + "src/app/(dashboard)/prompts/_components/prompt_editor_view/conversation_panel/MessageInput.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/prompts/components/prompt_editor_view/conversation_panel/index.tsx": { + "src/app/(dashboard)/prompts/_components/prompt_editor_view/conversation_panel/index.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/app/(dashboard)/prompts/components/prompt_editor_view/conversation_panel/useConversation.ts": { + "src/app/(dashboard)/prompts/_components/prompt_editor_view/conversation_panel/useConversation.ts": { "no-restricted-syntax": { "count": 1 } }, - "src/app/(dashboard)/prompts/components/prompt_info.tsx": { + "src/app/(dashboard)/prompts/_components/prompt_info.tsx": { + "no-nested-ternary": { + "count": 3 + }, "no-restricted-imports": { "count": 1 }, @@ -577,11 +655,22 @@ "count": 2 } }, - "src/app/(dashboard)/prompts/components/prompt_table.tsx": { + "src/app/(dashboard)/prompts/_components/prompt_table.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } }, + "src/app/(dashboard)/router-settings/_components/general_settings.tsx": { + "no-nested-ternary": { + "count": 3 + }, + "no-restricted-imports": { + "count": 2 + } + }, "src/app/(dashboard)/search-tools/_components/CreateSearchTools.tsx": { "no-restricted-imports": { "count": 1 @@ -635,6 +724,9 @@ } }, "src/app/(dashboard)/users/_components/user_edit_view.test.tsx": { + "no-nested-ternary": { + "count": 1 + }, "react/display-name": { "count": 1 } @@ -648,6 +740,9 @@ } }, "src/app/(dashboard)/users/_components/view_users.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 }, @@ -664,6 +759,9 @@ } }, "src/app/(dashboard)/users/_components/view_users/table.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } @@ -677,6 +775,9 @@ } }, "src/app/(dashboard)/workflows/WorkflowRuns.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-syntax": { "count": 3 }, @@ -684,6 +785,11 @@ "count": 1 } }, + "src/app/chat/page.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, "src/app/login/LoginPage.tsx": { "react-hooks/set-state-in-effect": { "count": 2 @@ -715,6 +821,9 @@ } }, "src/components/AIHub/ModelHubTable.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } @@ -746,6 +855,9 @@ } }, "src/components/AIHub/forms/MakeMCPPublicForm.tsx": { + "no-nested-ternary": { + "count": 2 + }, "no-restricted-imports": { "count": 1 }, @@ -761,14 +873,6 @@ "count": 1 } }, - "src/components/AdminPanel.tsx": { - "no-restricted-imports": { - "count": 1 - }, - "react-hooks/set-state-in-effect": { - "count": 1 - } - }, "src/components/CreateUserButton.tsx": { "no-restricted-imports": { "count": 1 @@ -778,11 +882,17 @@ } }, "src/components/DeletedKeysPage/DeletedKeysTable/DeletedKeysTable.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } }, "src/components/DeletedTeamsPage/DeletedTeamsTable/DeletedTeamsTable.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } @@ -815,11 +925,26 @@ "count": 1 } }, + "src/components/GuardrailSettingsView.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, + "src/components/GuardrailsMonitor/LogViewer.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, "src/components/HelpLink.test.tsx": { "unused-imports/no-unused-imports": { "count": 1 } }, + "src/components/ModelSelect/PaginatedModelSelect/PaginatedModelSelect.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, "src/components/Navbar/BlogDropdown/BlogDropdown.test.tsx": { "max-nested-callbacks": { "count": 12 @@ -836,6 +961,9 @@ } }, "src/components/OldTeams.tsx": { + "no-nested-ternary": { + "count": 4 + }, "no-restricted-imports": { "count": 1 }, @@ -856,7 +984,15 @@ "count": 1 } }, + "src/components/Settings/AdminSettings/HashicorpVault/HashicorpVault.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, "src/components/Settings/AdminSettings/MCPSemanticFilterSettings/MCPSemanticFilterSettings.tsx": { + "no-nested-ternary": { + "count": 1 + }, "react-hooks/set-state-in-effect": { "count": 1 } @@ -871,6 +1007,11 @@ "count": 1 } }, + "src/components/Settings/AdminSettings/SSOSettings/RedactableField.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, "src/components/Settings/AdminSettings/SSOSettings/SSOSettingsLoadingSkeleton.test.tsx": { "max-nested-callbacks": { "count": 4 @@ -881,7 +1022,15 @@ "count": 2 } }, + "src/components/Settings/AdminSettings/UISettings/UISettings.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, "src/components/Settings/LoggingAndAlerts/LoggingCallbacks/LoggingCallbacksTable.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } @@ -907,12 +1056,20 @@ "count": 1 } }, + "src/components/TeamSSOSettings.test.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, "src/components/ToolDetail.tsx": { "unused-imports/no-unused-imports": { "count": 2 } }, "src/components/ToolPolicies.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 }, @@ -932,6 +1089,9 @@ } }, "src/components/UsageIndicator.tsx": { + "no-nested-ternary": { + "count": 4 + }, "no-restricted-imports": { "count": 1 }, @@ -949,6 +1109,11 @@ "count": 1 } }, + "src/components/UsagePage/components/EndpointUsage/components/EndpointUsageTable.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, "src/components/UsagePage/components/EntityUsage/EntityUsage.tsx": { "no-restricted-imports": { "count": 1 @@ -975,11 +1140,17 @@ } }, "src/components/UsagePage/components/UsageAIChatPanel.tsx": { + "no-nested-ternary": { + "count": 1 + }, "react-hooks/immutability": { "count": 1 } }, "src/components/UsagePage/components/UsagePageView.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 }, @@ -999,16 +1170,25 @@ } }, "src/components/VirtualKeysPage/VirtualKeysTable.tsx": { + "no-nested-ternary": { + "count": 2 + }, "no-restricted-imports": { "count": 1 } }, "src/components/activity_metrics.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } }, "src/components/add_model/AddModelForm.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } @@ -1045,11 +1225,22 @@ } }, "src/components/add_model/litellm_model_name.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } }, + "src/components/add_model/model_connection_test.tsx": { + "no-nested-ternary": { + "count": 2 + } + }, "src/components/add_model/provider_specific_fields.tsx": { + "no-nested-ternary": { + "count": 5 + }, "no-restricted-imports": { "count": 1 }, @@ -1079,6 +1270,9 @@ } }, "src/components/agents/add_agent_form.tsx": { + "no-nested-ternary": { + "count": 3 + }, "no-restricted-imports": { "count": 1 }, @@ -1102,7 +1296,15 @@ "count": 1 } }, + "src/components/agents/agent_form_fields.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, "src/components/agents/agent_info.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 }, @@ -1110,7 +1312,20 @@ "count": 1 } }, + "src/components/agents/agent_virtual_keys.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, + "src/components/agents/dynamic_agent_form_fields.tsx": { + "no-nested-ternary": { + "count": 2 + } + }, "src/components/alerting/dynamic_form.tsx": { + "no-nested-ternary": { + "count": 4 + }, "no-restricted-imports": { "count": 1 } @@ -1123,6 +1338,31 @@ "count": 1 } }, + "src/components/chat/KeysPanel.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, + "src/components/chat/MCPAppsPanel.tsx": { + "no-nested-ternary": { + "count": 7 + } + }, + "src/components/chat/MCPConnectPicker.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, + "src/components/chat/MCPCredentialsTab.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, + "src/components/chat/UsagePanel.tsx": { + "no-nested-ternary": { + "count": 2 + } + }, "src/components/claude_code_plugins.tsx": { "no-restricted-imports": { "count": 1 @@ -1145,6 +1385,9 @@ } }, "src/components/claude_code_plugins/plugin_table.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } @@ -1235,6 +1478,9 @@ } }, "src/components/common_components/chartUtils.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } @@ -1250,6 +1496,9 @@ } }, "src/components/common_components/simple_table.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } @@ -1285,11 +1534,6 @@ "count": 1 } }, - "src/components/general_settings.tsx": { - "no-restricted-imports": { - "count": 2 - } - }, "src/components/guardrails.tsx": { "react-hooks/set-state-in-effect": { "count": 1 @@ -1300,17 +1544,28 @@ "count": 1 } }, + "src/components/guardrails/GuardrailTestPlayground.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, "src/components/guardrails/GuardrailTestResults.tsx": { "no-restricted-imports": { "count": 1 } }, "src/components/guardrails/TeamGuardrailsTab.tsx": { + "no-nested-ternary": { + "count": 2 + }, "react-hooks/set-state-in-effect": { "count": 1 } }, "src/components/guardrails/add_guardrail_form.tsx": { + "no-nested-ternary": { + "count": 4 + }, "react-hooks/set-state-in-effect": { "count": 1 }, @@ -1319,11 +1574,17 @@ } }, "src/components/guardrails/content_filter/CompetitorIntentConfiguration.tsx": { + "no-nested-ternary": { + "count": 1 + }, "react-hooks/set-state-in-effect": { "count": 1 } }, "src/components/guardrails/content_filter/ContentCategoryConfiguration.tsx": { + "no-nested-ternary": { + "count": 3 + }, "react-hooks/set-state-in-effect": { "count": 1 } @@ -1342,6 +1603,9 @@ } }, "src/components/guardrails/custom_code/CustomCodeModal.tsx": { + "no-nested-ternary": { + "count": 6 + }, "no-restricted-imports": { "count": 1 }, @@ -1372,18 +1636,27 @@ } }, "src/components/guardrails/guardrail_optional_params.tsx": { + "no-nested-ternary": { + "count": 5 + }, "react-hooks/set-state-in-effect": { "count": 1 } }, "src/components/guardrails/guardrail_provider_fields.tsx": { + "no-nested-ternary": { + "count": 5 + }, "react-hooks/set-state-in-effect": { "count": 1 } }, "src/components/guardrails/guardrail_table.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { - "count": 2 + "count": 1 } }, "src/components/guardrails/tool_permission/ToolPermissionRulesEditor.tsx": { @@ -1407,6 +1680,9 @@ "src/components/llm_calls/chat_completion.tsx": { "max-params": { "count": 1 + }, + "no-nested-ternary": { + "count": 1 } }, "src/components/llm_calls/responses_api.tsx": { @@ -1444,7 +1720,15 @@ "count": 1 } }, + "src/components/mcp_tools/MCPToolArgumentsForm.tsx": { + "no-nested-ternary": { + "count": 5 + } + }, "src/components/mcp_tools/MCPToolsetsTab.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 }, @@ -1456,11 +1740,17 @@ } }, "src/components/mcp_tools/McpCrudPermissionPanel.tsx": { + "no-nested-ternary": { + "count": 3 + }, "no-restricted-imports": { "count": 1 } }, "src/components/mcp_tools/OAuthFormFields.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } @@ -1471,6 +1761,9 @@ } }, "src/components/mcp_tools/ToolTestPanel.tsx": { + "no-nested-ternary": { + "count": 3 + }, "no-restricted-imports": { "count": 1 }, @@ -1478,12 +1771,20 @@ "count": 1 } }, + "src/components/mcp_tools/UserEnvVarsModal.tsx": { + "no-nested-ternary": { + "count": 2 + } + }, "src/components/mcp_tools/create_mcp_server.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 }, "react-hooks/set-state-in-effect": { - "count": 5 + "count": 4 } }, "src/components/mcp_tools/mcp_connect.tsx": { @@ -1495,6 +1796,9 @@ } }, "src/components/mcp_tools/mcp_connection_status.tsx": { + "no-nested-ternary": { + "count": 3 + }, "no-restricted-imports": { "count": 1 } @@ -1515,6 +1819,9 @@ } }, "src/components/mcp_tools/mcp_server_edit.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 }, @@ -1531,6 +1838,9 @@ } }, "src/components/mcp_tools/mcp_servers.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 }, @@ -1544,6 +1854,9 @@ } }, "src/components/mcp_tools/mcp_tools.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 }, @@ -1575,6 +1888,9 @@ } }, "src/components/model_dashboard/HealthCheckComponent.tsx": { + "no-nested-ternary": { + "count": 3 + }, "no-restricted-imports": { "count": 1 }, @@ -1583,6 +1899,9 @@ } }, "src/components/model_dashboard/all_models_table.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } @@ -1591,11 +1910,17 @@ "max-params": { "count": 1 }, + "no-nested-ternary": { + "count": 2 + }, "no-restricted-imports": { "count": 1 } }, "src/components/model_dashboard/table.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } @@ -1619,6 +1944,9 @@ } }, "src/components/model_info_view.tsx": { + "no-nested-ternary": { + "count": 14 + }, "no-restricted-imports": { "count": 1 }, @@ -1627,6 +1955,9 @@ } }, "src/components/molecules/filter.tsx": { + "no-nested-ternary": { + "count": 2 + }, "react-hooks/use-memo": { "count": 1 } @@ -1643,6 +1974,9 @@ "max-params": { "count": 1 }, + "no-nested-ternary": { + "count": 2 + }, "no-restricted-imports": { "count": 1 } @@ -1656,6 +1990,9 @@ "max-params": { "count": 23 }, + "no-nested-ternary": { + "count": 5 + }, "no-restricted-syntax": { "count": 154 } @@ -1697,11 +2034,6 @@ "count": 1 } }, - "src/components/organizations.tsx": { - "no-restricted-imports": { - "count": 1 - } - }, "src/components/page_utils.test.ts": { "max-nested-callbacks": { "count": 3 @@ -1731,6 +2063,9 @@ } }, "src/components/permissions/MCPServerPermissions.tsx": { + "no-nested-ternary": { + "count": 3 + }, "no-restricted-imports": { "count": 1 } @@ -1740,7 +2075,12 @@ "count": 1 } }, - "src/components/policies/add_attachment_form.tsx": { + "src/components/policies/PolicySelector.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, + "src/app/(dashboard)/policies/_components/add_attachment_form.tsx": { "no-restricted-imports": { "count": 1 }, @@ -1748,7 +2088,7 @@ "count": 1 } }, - "src/components/policies/add_policy_form.tsx": { + "src/app/(dashboard)/policies/_components/add_policy_form.tsx": { "no-restricted-imports": { "count": 1 }, @@ -1759,7 +2099,10 @@ "count": 1 } }, - "src/components/policies/ai_suggestion_modal.tsx": { + "src/app/(dashboard)/policies/_components/ai_suggestion_modal.tsx": { + "no-nested-ternary": { + "count": 10 + }, "no-restricted-imports": { "count": 1 }, @@ -1767,37 +2110,46 @@ "count": 1 } }, - "src/components/policies/attachment_table.test.tsx": { + "src/app/(dashboard)/policies/_components/attachment_table.test.tsx": { "react/display-name": { "count": 1 } }, - "src/components/policies/attachment_table.tsx": { + "src/app/(dashboard)/policies/_components/attachment_table.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } }, - "src/components/policies/guardrail_selection_modal.tsx": { + "src/app/(dashboard)/policies/_components/guardrail_selection_modal.tsx": { + "no-nested-ternary": { + "count": 1 + }, "react-hooks/set-state-in-effect": { "count": 1 } }, - "src/components/policies/impact_popover.test.tsx": { + "src/app/(dashboard)/policies/_components/impact_popover.test.tsx": { "react/display-name": { "count": 1 } }, - "src/components/policies/impact_popover.tsx": { + "src/app/(dashboard)/policies/_components/impact_popover.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } }, - "src/components/policies/index.test.tsx": { + "src/app/(dashboard)/policies/_components/index.test.tsx": { "react/display-name": { "count": 1 } }, - "src/components/policies/index.tsx": { + "src/app/(dashboard)/policies/_components/index.tsx": { "no-restricted-imports": { "count": 1 }, @@ -1805,7 +2157,10 @@ "count": 1 } }, - "src/components/policies/pipeline_flow_builder.tsx": { + "src/app/(dashboard)/policies/_components/pipeline_flow_builder.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 }, @@ -1813,7 +2168,7 @@ "count": 2 } }, - "src/components/policies/policy_info.tsx": { + "src/app/(dashboard)/policies/_components/policy_info.tsx": { "no-restricted-imports": { "count": 1 }, @@ -1821,17 +2176,20 @@ "count": 1 } }, - "src/components/policies/policy_table.test.tsx": { + "src/app/(dashboard)/policies/_components/policy_table.test.tsx": { "react/display-name": { "count": 1 } }, - "src/components/policies/policy_table.tsx": { + "src/app/(dashboard)/policies/_components/policy_table.tsx": { + "no-nested-ternary": { + "count": 2 + }, "no-restricted-imports": { "count": 1 } }, - "src/components/policies/policy_test_panel.tsx": { + "src/app/(dashboard)/policies/_components/policy_test_panel.tsx": { "no-restricted-imports": { "count": 1 }, @@ -1839,7 +2197,7 @@ "count": 1 } }, - "src/components/policies/template_parameter_modal.tsx": { + "src/app/(dashboard)/policies/_components/template_parameter_modal.tsx": { "no-restricted-imports": { "count": 1 }, @@ -1856,6 +2214,9 @@ } }, "src/components/public_model_hub.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } @@ -1865,12 +2226,25 @@ "count": 1 } }, + "src/components/router_settings/ReliabilityRetriesSection.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, "src/components/routing_groups/index.tsx": { "react-hooks/preserve-manual-memoization": { "count": 1 } }, + "src/components/search_tools/SearchToolSelector.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, "src/components/settings.tsx": { + "no-nested-ternary": { + "count": 2 + }, "no-restricted-imports": { "count": 1 } @@ -1898,17 +2272,17 @@ "count": 1 } }, - "src/components/tag_management/TagTable.tsx": { + "src/app/(dashboard)/tag-management/_components/TagTable.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/components/tag_management/components/CreateTagModal.tsx": { + "src/app/(dashboard)/tag-management/_components/components/CreateTagModal.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/components/tag_management/index.tsx": { + "src/app/(dashboard)/tag-management/_components/index.tsx": { "no-restricted-imports": { "count": 1 }, @@ -1916,7 +2290,7 @@ "count": 1 } }, - "src/components/tag_management/tag_info.tsx": { + "src/app/(dashboard)/tag-management/_components/tag_info.tsx": { "no-restricted-imports": { "count": 1 }, @@ -1925,6 +2299,9 @@ } }, "src/components/team/EditMembership.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 } @@ -1935,6 +2312,9 @@ } }, "src/components/team/TeamInfo.tsx": { + "no-nested-ternary": { + "count": 3 + }, "no-restricted-imports": { "count": 1 }, @@ -1943,6 +2323,9 @@ } }, "src/components/team/TeamVirtualKeysTable.tsx": { + "no-nested-ternary": { + "count": 2 + }, "no-restricted-imports": { "count": 1 } @@ -1966,6 +2349,9 @@ } }, "src/components/templates/key_edit_view.tsx": { + "no-nested-ternary": { + "count": 2 + }, "no-restricted-imports": { "count": 1 } @@ -1976,6 +2362,9 @@ } }, "src/components/templates/key_info_view.tsx": { + "no-nested-ternary": { + "count": 1 + }, "no-restricted-imports": { "count": 1 }, @@ -1983,17 +2372,6 @@ "count": 1 } }, - "src/components/usage.tsx": { - "no-restricted-imports": { - "count": 2 - }, - "react-hooks/immutability": { - "count": 1 - }, - "react-hooks/purity": { - "count": 1 - } - }, "src/components/user_agent_activity.tsx": { "no-restricted-imports": { "count": 2 @@ -2010,12 +2388,15 @@ "count": 2 } }, - "src/components/vector_store_management/CreateVectorStore.tsx": { + "src/app/(dashboard)/vector-stores/_components/CreateVectorStore.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/components/vector_store_management/VectorStoreForm.tsx": { + "src/app/(dashboard)/vector-stores/_components/VectorStoreForm.tsx": { + "no-nested-ternary": { + "count": 2 + }, "no-restricted-imports": { "count": 1 }, @@ -2023,12 +2404,12 @@ "count": 1 } }, - "src/components/vector_store_management/VectorStoreTable.tsx": { + "src/app/(dashboard)/vector-stores/_components/VectorStoreTable.tsx": { "no-restricted-imports": { "count": 1 } }, - "src/components/vector_store_management/index.tsx": { + "src/app/(dashboard)/vector-stores/_components/index.tsx": { "no-restricted-imports": { "count": 1 }, @@ -2036,7 +2417,7 @@ "count": 1 } }, - "src/components/vector_store_management/vector_store_info.tsx": { + "src/app/(dashboard)/vector-stores/_components/vector_store_info.tsx": { "no-restricted-imports": { "count": 1 }, @@ -2044,12 +2425,38 @@ "count": 1 } }, + "src/components/view_logs/EvalViewer/EvalViewer.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, "src/components/view_logs/GuardrailViewer/CompliancePanel.tsx": { + "no-nested-ternary": { + "count": 2 + }, "react-hooks/set-state-in-effect": { "count": 1 } }, + "src/components/view_logs/GuardrailViewer/ContentFilterDetails.tsx": { + "no-nested-ternary": { + "count": 1 + } + }, + "src/components/view_logs/GuardrailViewer/GuardrailViewer.tsx": { + "no-nested-ternary": { + "count": 4 + } + }, + "src/components/view_logs/LogDetailsDrawer/LogDetailContent.tsx": { + "no-nested-ternary": { + "count": 4 + } + }, "src/components/view_logs/LogDetailsDrawer/LogDetailsDrawer.tsx": { + "no-nested-ternary": { + "count": 3 + }, "react-hooks/set-state-in-effect": { "count": 2 } @@ -2059,14 +2466,19 @@ "count": 2 } }, + "src/components/view_logs/LogDetailsDrawer/prettyMessagesUtils.ts": { + "no-nested-ternary": { + "count": 1 + } + }, "src/components/view_logs/LogDetailsDrawer/useKeyboardNavigation.ts": { "react-hooks/immutability": { "count": 2 } }, - "src/components/view_logs/columns.tsx": { - "no-restricted-imports": { - "count": 1 + "src/components/view_logs/LogsTableToolbar.tsx": { + "no-nested-ternary": { + "count": 4 } }, "src/components/view_logs/index.tsx": { @@ -2078,8 +2490,8 @@ } }, "src/components/view_logs/table.tsx": { - "no-restricted-imports": { - "count": 1 + "no-nested-ternary": { + "count": 2 } }, "src/components/view_user_spend.tsx": { @@ -2118,6 +2530,9 @@ } }, "src/hooks/useTestMCPConnection.tsx": { + "no-nested-ternary": { + "count": 1 + }, "react-hooks/set-state-in-effect": { "count": 1 } @@ -2135,6 +2550,11 @@ "count": 1 } }, + "src/lib/http/client.ts": { + "no-nested-ternary": { + "count": 1 + } + }, "src/utils/dataUtils.test.ts": { "max-nested-callbacks": { "count": 1 diff --git a/ui/litellm-dashboard/eslint.config.mjs b/ui/litellm-dashboard/eslint.config.mjs index acdd0c91309..0cf5b4ff655 100644 --- a/ui/litellm-dashboard/eslint.config.mjs +++ b/ui/litellm-dashboard/eslint.config.mjs @@ -3,6 +3,7 @@ import tseslint from "typescript-eslint"; import nextCoreWebVitals from "eslint-config-next/core-web-vitals"; import prettier from "eslint-config-prettier/flat"; import unusedImports from "eslint-plugin-unused-imports"; +import local from "./scripts/eslint-rules/index.mjs"; const eslintConfig = [ { @@ -13,9 +14,11 @@ const eslintConfig = [ ...nextCoreWebVitals, prettier, { - plugins: { "unused-imports": unusedImports }, + plugins: { "unused-imports": unusedImports, local }, rules: { "unused-imports/no-unused-imports": "error", + "local/no-large-inline-object-arg": "warn", + "local/no-long-condition-chain": "warn", "@typescript-eslint/no-explicit-any": "warn", "no-console": ["warn", { allow: ["warn", "error"] }], "@typescript-eslint/no-unused-vars": "off", @@ -28,6 +31,7 @@ const eslintConfig = [ "no-useless-escape": "off", "no-self-assign": "error", "no-var": "error", + "no-nested-ternary": "error", "react/no-danger": "error", complexity: ["warn", 20], "max-depth": ["warn", 4], diff --git a/ui/litellm-dashboard/package-lock.json b/ui/litellm-dashboard/package-lock.json index ae3660f59e9..56c0a4f9500 100644 --- a/ui/litellm-dashboard/package-lock.json +++ b/ui/litellm-dashboard/package-lock.json @@ -34,6 +34,7 @@ "react-json-view-lite": "2.5.0", "react-markdown": "9.1.0", "react-syntax-highlighter": "15.6.6", + "recharts": "3.9.2", "remark-gfm": "4.0.1", "tailwind-merge": "3.4.0", "uuid": "14.0.0" @@ -2927,6 +2928,32 @@ "npm": ">=9.5.0" } }, + "node_modules/@reduxjs/toolkit": { + "version": "2.12.0", + "resolved": "https://registry.npmjs.org/@reduxjs/toolkit/-/toolkit-2.12.0.tgz", + "integrity": "sha512-KiT+RzZbp6mQET+Mg+h2c97+9j1sNflUxQkIHI7Yuzf6Peu+OYpmkn6nbHWmLLWj+1ZODUJFwGZ7gx3L9R9EOw==", + "license": "MIT", + "dependencies": { + "@standard-schema/spec": "^1.0.0", + "@standard-schema/utils": "^0.3.0", + "immer": "^11.0.0", + "redux": "^5.0.1", + "redux-thunk": "^3.1.0", + "reselect": "^5.1.0" + }, + "peerDependencies": { + "react": "^16.9.0 || ^17.0.0 || ^18 || ^19", + "react-redux": "^7.2.1 || ^8.1.3 || ^9.0.0" + }, + "peerDependenciesMeta": { + "react": { + "optional": true + }, + "react-redux": { + "optional": true + } + } + }, "node_modules/@rollup/rollup-android-arm-eabi": { "version": "4.61.1", "resolved": "https://registry.npmjs.org/@rollup/rollup-android-arm-eabi/-/rollup-android-arm-eabi-4.61.1.tgz", @@ -3284,6 +3311,18 @@ "dev": true, "license": "MIT" }, + "node_modules/@standard-schema/spec": { + "version": "1.1.0", + "resolved": "https://registry.npmjs.org/@standard-schema/spec/-/spec-1.1.0.tgz", + "integrity": "sha512-l2aFy5jALhniG5HgqrD6jXLi/rUWrKvqN/qJx6yoJsgKhblVd+iqqU4RCXavm/jPityDo5TCvKMnpjKnOriy0w==", + "license": "MIT" + }, + "node_modules/@standard-schema/utils": { + "version": "0.3.0", + "resolved": "https://registry.npmjs.org/@standard-schema/utils/-/utils-0.3.0.tgz", + "integrity": "sha512-e7Mew686owMaPJVNNLs55PUvgz371nKgwsc4vxE49zsODpJEnxgxRo2y/OKrqueavXgZNMDVj3DdHFlaSAeU8g==", + "license": "MIT" + }, "node_modules/@swc/helpers": { "version": "0.5.15", "resolved": "https://registry.npmjs.org/@swc/helpers/-/helpers-0.5.15.tgz", @@ -3804,6 +3843,42 @@ "react-dom": ">=16.6.0" } }, + "node_modules/@tremor/react/node_modules/eventemitter3": { + "version": "4.0.7", + "resolved": "https://registry.npmjs.org/eventemitter3/-/eventemitter3-4.0.7.tgz", + "integrity": "sha512-8guHBZCwKnFhYdHr2ysuRWErTwhoN2X8XELRlrRwpmfeY2jjuUN4taQMsULKUVo1K4DvZl+0pgfyoysHxvmvEw==", + "license": "MIT" + }, + "node_modules/@tremor/react/node_modules/react-is": { + "version": "18.3.1", + "resolved": "https://registry.npmjs.org/react-is/-/react-is-18.3.1.tgz", + "integrity": "sha512-/LLMVyas0ljjAtoYiPqYiL8VWXzUUdThrmU5+n20DZv+a+ClRoevUzw5JxU+Ieh5/c87ytoTBV9G1FiKfNJdmg==", + "license": "MIT" + }, + "node_modules/@tremor/react/node_modules/recharts": { + "version": "2.15.4", + "resolved": "https://registry.npmjs.org/recharts/-/recharts-2.15.4.tgz", + "integrity": "sha512-UT/q6fwS3c1dHbXv2uFgYJ9BMFHu3fwnd7AYZaEQhXuYQ4hgsxLvsUXzGdKeZrW5xopzDCvuA2N41WJ88I7zIw==", + "deprecated": "1.x and 2.x branches are no longer active. Bump to Recharts v3 to receive latest features and bugfixes. See https://github.com/recharts/recharts/wiki/3.0-migration-guide", + "license": "MIT", + "dependencies": { + "clsx": "^2.0.0", + "eventemitter3": "^4.0.1", + "lodash": "^4.17.21", + "react-is": "^18.3.1", + "react-smooth": "^4.0.4", + "recharts-scale": "^0.4.4", + "tiny-invariant": "^1.3.1", + "victory-vendor": "^36.6.8" + }, + "engines": { + "node": ">=14" + }, + "peerDependencies": { + "react": "^16.0.0 || ^17.0.0 || ^18.0.0 || ^19.0.0", + "react-dom": "^16.0.0 || ^17.0.0 || ^18.0.0 || ^19.0.0" + } + }, "node_modules/@tremor/react/node_modules/tailwind-merge": { "version": "2.6.1", "resolved": "https://registry.npmjs.org/tailwind-merge/-/tailwind-merge-2.6.1.tgz", @@ -3814,6 +3889,28 @@ "url": "https://github.com/sponsors/dcastil" } }, + "node_modules/@tremor/react/node_modules/victory-vendor": { + "version": "36.9.2", + "resolved": "https://registry.npmjs.org/victory-vendor/-/victory-vendor-36.9.2.tgz", + "integrity": "sha512-PnpQQMuxlwYdocC8fIJqVXvkeViHYzotI+NJrCuav0ZYFoq912ZHBk3mCeuj+5/VpodOjPe1z0Fk2ihgzlXqjQ==", + "license": "MIT AND ISC", + "dependencies": { + "@types/d3-array": "^3.0.3", + "@types/d3-ease": "^3.0.0", + "@types/d3-interpolate": "^3.0.1", + "@types/d3-scale": "^4.0.2", + "@types/d3-shape": "^3.1.0", + "@types/d3-time": "^3.0.0", + "@types/d3-timer": "^3.0.0", + "d3-array": "^3.1.6", + "d3-ease": "^3.0.1", + "d3-interpolate": "^3.0.1", + "d3-scale": "^4.0.2", + "d3-shape": "^3.1.0", + "d3-time": "^3.0.0", + "d3-timer": "^3.0.1" + } + }, "node_modules/@tybys/wasm-util": { "version": "0.10.3", "resolved": "https://registry.npmjs.org/@tybys/wasm-util/-/wasm-util-0.10.3.tgz", @@ -4069,6 +4166,12 @@ "integrity": "sha512-ko/gIFJRv177XgZsZcBwnqJN5x/Gien8qNOn0D5bQU/zAzVf9Zt3BlcUiLqhV9y4ARk0GbT3tnUiPNgnTXzc/Q==", "license": "MIT" }, + "node_modules/@types/use-sync-external-store": { + "version": "0.0.6", + "resolved": "https://registry.npmjs.org/@types/use-sync-external-store/-/use-sync-external-store-0.0.6.tgz", + "integrity": "sha512-zFDAD+tlpf2r4asuHEj0XH6pY6i0g5NeAHPn+15wk3BV6JA69eERFXC1gyGThDkVa1zCyKr5jox1+2LbV/AMLg==", + "license": "MIT" + }, "node_modules/@typescript-eslint/eslint-plugin": { "version": "8.60.1", "resolved": "https://registry.npmjs.org/@typescript-eslint/eslint-plugin/-/eslint-plugin-8.60.1.tgz", @@ -6309,6 +6412,16 @@ "url": "https://github.com/sponsors/ljharb" } }, + "node_modules/es-toolkit": { + "version": "1.49.0", + "resolved": "https://registry.npmjs.org/es-toolkit/-/es-toolkit-1.49.0.tgz", + "integrity": "sha512-G5iZ6Pc/FNRY/soKZHC+TxGDD83rHUDXxzaWhGCX44vAv/tMs56WMusnm/KMNK+luUPsgA9U28cGr4RDlSzL2g==", + "license": "MIT", + "workspaces": [ + "docs", + "benchmarks" + ] + }, "node_modules/esbuild": { "version": "0.28.1", "resolved": "https://registry.npmjs.org/esbuild/-/esbuild-0.28.1.tgz", @@ -6872,9 +6985,9 @@ } }, "node_modules/eventemitter3": { - "version": "4.0.7", - "resolved": "https://registry.npmjs.org/eventemitter3/-/eventemitter3-4.0.7.tgz", - "integrity": "sha512-8guHBZCwKnFhYdHr2ysuRWErTwhoN2X8XELRlrRwpmfeY2jjuUN4taQMsULKUVo1K4DvZl+0pgfyoysHxvmvEw==", + "version": "5.0.4", + "resolved": "https://registry.npmjs.org/eventemitter3/-/eventemitter3-5.0.4.tgz", + "integrity": "sha512-mlsTRyGaPBjPedk6Bvw+aqbsXDtoAyAzm5MO7JgU+yVRyMQ5O8bD4Kcci7BS85f93veegeCPkL8R4GLClnjLFw==", "license": "MIT" }, "node_modules/expect-type": { @@ -6901,9 +7014,9 @@ "license": "MIT" }, "node_modules/fast-equals": { - "version": "5.4.0", - "resolved": "https://registry.npmjs.org/fast-equals/-/fast-equals-5.4.0.tgz", - "integrity": "sha512-jt2DW/aNFNwke7AUd+Z+e6pz39KO5rzdbbFCg2sGafS4mk13MI7Z8O5z9cADNn5lhGODIgLwug6TZO2ctf7kcw==", + "version": "5.4.1", + "resolved": "https://registry.npmjs.org/fast-equals/-/fast-equals-5.4.1.tgz", + "integrity": "sha512-DjlFSM5Pk9cGcL0q5QXl66eGzx0N6szNgaswwc5ZphlBohjTVJSnGgI+rJVOgOi65qUoQnDZN4nDqi33udtydQ==", "license": "MIT", "engines": { "node": ">=6.0.0" @@ -7688,6 +7801,16 @@ "node": ">= 4" } }, + "node_modules/immer": { + "version": "11.1.11", + "resolved": "https://registry.npmjs.org/immer/-/immer-11.1.11.tgz", + "integrity": "sha512-qzXuyXAkPySAGYkfsAwodDPWT8Zm7/Uo5BNt4BjhMhG5WlWyZZ4wQqnWwdS8kjlQ1Cwu6gjw3A6+0gTQwlyYtw==", + "license": "MIT", + "funding": { + "type": "opencollective", + "url": "https://opencollective.com/immer" + } + }, "node_modules/import-fresh": { "version": "3.3.1", "resolved": "https://registry.npmjs.org/import-fresh/-/import-fresh-3.3.1.tgz", @@ -11589,7 +11712,6 @@ "version": "17.0.2", "resolved": "https://registry.npmjs.org/react-is/-/react-is-17.0.2.tgz", "integrity": "sha512-w2GsyukL62IJnlaff/nRegPQR94C/XXamvMWmSHRJ4y7Ts/4ocGRmTHvOs8PSE6pB3dWOrD/nueuU5sduBsQ4w==", - "dev": true, "license": "MIT" }, "node_modules/react-json-view-lite": { @@ -11631,6 +11753,29 @@ "react": ">=18" } }, + "node_modules/react-redux": { + "version": "9.3.0", + "resolved": "https://registry.npmjs.org/react-redux/-/react-redux-9.3.0.tgz", + "integrity": "sha512-KQopgqFo/p/fgmAs5qz6p5RWaNAzq40WAu7fJIXnQpYxFPbJYtsJPWvGeF2rOBaY/kEuV77AVsX8TsQzKm+A/g==", + "license": "MIT", + "dependencies": { + "@types/use-sync-external-store": "^0.0.6", + "use-sync-external-store": "^1.4.0" + }, + "peerDependencies": { + "@types/react": "^18.2.25 || ^19", + "react": "^18.0 || ^19", + "redux": "^5.0.0" + }, + "peerDependenciesMeta": { + "@types/react": { + "optional": true + }, + "redux": { + "optional": true + } + } + }, "node_modules/react-smooth": { "version": "4.0.4", "resolved": "https://registry.npmjs.org/react-smooth/-/react-smooth-4.0.4.tgz", @@ -11707,26 +11852,33 @@ } }, "node_modules/recharts": { - "version": "2.15.4", - "resolved": "https://registry.npmjs.org/recharts/-/recharts-2.15.4.tgz", - "integrity": "sha512-UT/q6fwS3c1dHbXv2uFgYJ9BMFHu3fwnd7AYZaEQhXuYQ4hgsxLvsUXzGdKeZrW5xopzDCvuA2N41WJ88I7zIw==", + "version": "3.9.2", + "resolved": "https://registry.npmjs.org/recharts/-/recharts-3.9.2.tgz", + "integrity": "sha512-G4fy+Pk46RaXgwWMh+Nzhyo/lbFAVqXo9gtetlyehe6Ehge9CsgDuOTwQDD+i1+llaLktNBiNq4bhnGlDRXFtw==", "license": "MIT", + "workspaces": [ + "www" + ], "dependencies": { - "clsx": "^2.0.0", - "eventemitter3": "^4.0.1", - "lodash": "^4.17.21", - "react-is": "^18.3.1", - "react-smooth": "^4.0.4", - "recharts-scale": "^0.4.4", - "tiny-invariant": "^1.3.1", - "victory-vendor": "^36.6.8" + "@reduxjs/toolkit": "^1.9.0 || 2.x.x", + "clsx": "^2.1.1", + "decimal.js-light": "^2.5.1", + "es-toolkit": "^1.39.3", + "eventemitter3": "^5.0.1", + "immer": "^11.1.8", + "react-redux": "8.x.x || 9.x.x", + "reselect": "5.2.0", + "tiny-invariant": "^1.3.3", + "use-sync-external-store": "^1.2.2", + "victory-vendor": "^37.0.2" }, "engines": { - "node": ">=14" + "node": ">=18" }, "peerDependencies": { - "react": "^16.0.0 || ^17.0.0 || ^18.0.0 || ^19.0.0", - "react-dom": "^16.0.0 || ^17.0.0 || ^18.0.0 || ^19.0.0" + "react": "^16.8.0 || ^17.0.0 || ^18.0.0 || ^19.0.0", + "react-dom": "^16.0.0 || ^17.0.0 || ^18.0.0 || ^19.0.0", + "react-is": "^16.8.0 || ^17.0.0 || ^18.0.0 || ^19.0.0" } }, "node_modules/recharts-scale": { @@ -11738,12 +11890,6 @@ "decimal.js-light": "^2.4.1" } }, - "node_modules/recharts/node_modules/react-is": { - "version": "18.3.1", - "resolved": "https://registry.npmjs.org/react-is/-/react-is-18.3.1.tgz", - "integrity": "sha512-/LLMVyas0ljjAtoYiPqYiL8VWXzUUdThrmU5+n20DZv+a+ClRoevUzw5JxU+Ieh5/c87ytoTBV9G1FiKfNJdmg==", - "license": "MIT" - }, "node_modules/redent": { "version": "3.0.0", "resolved": "https://registry.npmjs.org/redent/-/redent-3.0.0.tgz", @@ -11758,6 +11904,21 @@ "node": ">=8" } }, + "node_modules/redux": { + "version": "5.0.1", + "resolved": "https://registry.npmjs.org/redux/-/redux-5.0.1.tgz", + "integrity": "sha512-M9/ELqF6fy8FwmkpnF0S3YKOqMyoWJ4+CS5Efg2ct3oY9daQvd/Pc71FpGZsVsbl3Cpb+IIcjBDUnnyBdQbq4w==", + "license": "MIT" + }, + "node_modules/redux-thunk": { + "version": "3.1.0", + "resolved": "https://registry.npmjs.org/redux-thunk/-/redux-thunk-3.1.0.tgz", + "integrity": "sha512-NW2r5T6ksUKXCabzhL9z+h206HQw/NJkcLm1GPImRQ8IzfXwRGqjVhKJGauHirT0DAuyy6hjdnMZaRoAcy0Klw==", + "license": "MIT", + "peerDependencies": { + "redux": "^5.0.0" + } + }, "node_modules/reflect.getprototypeof": { "version": "1.0.10", "resolved": "https://registry.npmjs.org/reflect.getprototypeof/-/reflect.getprototypeof-1.0.10.tgz", @@ -13429,9 +13590,9 @@ } }, "node_modules/victory-vendor": { - "version": "36.9.2", - "resolved": "https://registry.npmjs.org/victory-vendor/-/victory-vendor-36.9.2.tgz", - "integrity": "sha512-PnpQQMuxlwYdocC8fIJqVXvkeViHYzotI+NJrCuav0ZYFoq912ZHBk3mCeuj+5/VpodOjPe1z0Fk2ihgzlXqjQ==", + "version": "37.3.6", + "resolved": "https://registry.npmjs.org/victory-vendor/-/victory-vendor-37.3.6.tgz", + "integrity": "sha512-SbPDPdDBYp+5MJHhBCAyI7wKM3d5ivekigc2Dk2s7pgbZ9wIgIBYGVw4zGHBml/qTFbexrofXW6Gu4noGxrOwQ==", "license": "MIT AND ISC", "dependencies": { "@types/d3-array": "^3.0.3", diff --git a/ui/litellm-dashboard/package.json b/ui/litellm-dashboard/package.json index 1b0ce315e4d..1747a40da56 100644 --- a/ui/litellm-dashboard/package.json +++ b/ui/litellm-dashboard/package.json @@ -50,6 +50,7 @@ "react-json-view-lite": "2.5.0", "react-markdown": "9.1.0", "react-syntax-highlighter": "15.6.6", + "recharts": "3.9.2", "remark-gfm": "4.0.1", "tailwind-merge": "3.4.0", "uuid": "14.0.0" diff --git a/ui/litellm-dashboard/scripts/eslint-rules/index.mjs b/ui/litellm-dashboard/scripts/eslint-rules/index.mjs new file mode 100644 index 00000000000..150ba1d02e9 --- /dev/null +++ b/ui/litellm-dashboard/scripts/eslint-rules/index.mjs @@ -0,0 +1,11 @@ +import noLargeInlineObjectArg from "./no-large-inline-object-arg.mjs"; +import noLongConditionChain from "./no-long-condition-chain.mjs"; + +const plugin = { + rules: { + "no-large-inline-object-arg": noLargeInlineObjectArg, + "no-long-condition-chain": noLongConditionChain, + }, +}; + +export default plugin; diff --git a/ui/litellm-dashboard/scripts/eslint-rules/no-large-inline-object-arg.mjs b/ui/litellm-dashboard/scripts/eslint-rules/no-large-inline-object-arg.mjs new file mode 100644 index 00000000000..5c5ae170e23 --- /dev/null +++ b/ui/litellm-dashboard/scripts/eslint-rules/no-large-inline-object-arg.mjs @@ -0,0 +1,41 @@ +const DEFAULT_MIN_PROPERTIES = 4; + +const isArgumentOf = (node) => { + const parent = node.parent; + if (parent == null) return false; + if (parent.type !== "CallExpression" && parent.type !== "NewExpression") return false; + return parent.arguments.includes(node); +}; + +const rule = { + meta: { + type: "suggestion", + docs: { + description: + "Disallow passing a large object literal inline as a call argument; assign it to a named variable first.", + }, + schema: [ + { + type: "object", + properties: { minProperties: { type: "integer", minimum: 1 } }, + additionalProperties: false, + }, + ], + messages: { + tooLarge: + "Object literal with {{count}} properties passed inline as an argument; assign it to a named variable first.", + }, + }, + create(context) { + const minProperties = context.options[0]?.minProperties ?? DEFAULT_MIN_PROPERTIES; + return { + ObjectExpression(node) { + if (!isArgumentOf(node)) return; + if (node.properties.length < minProperties) return; + context.report({ node, messageId: "tooLarge", data: { count: node.properties.length } }); + }, + }; + }, +}; + +export default rule; diff --git a/ui/litellm-dashboard/scripts/eslint-rules/no-long-condition-chain.mjs b/ui/litellm-dashboard/scripts/eslint-rules/no-long-condition-chain.mjs new file mode 100644 index 00000000000..638e57442e2 --- /dev/null +++ b/ui/litellm-dashboard/scripts/eslint-rules/no-long-condition-chain.mjs @@ -0,0 +1,41 @@ +const DEFAULT_MIN_CONDITIONS = 4; + +const isBooleanLogical = (node) => + node?.type === "LogicalExpression" && (node.operator === "&&" || node.operator === "||"); + +const countConditions = (node) => + isBooleanLogical(node) ? countConditions(node.left) + countConditions(node.right) : 1; + +const rule = { + meta: { + type: "suggestion", + docs: { + description: + "Disallow logical expressions that combine many conditions; extract the condition into a named boolean.", + }, + schema: [ + { + type: "object", + properties: { minConditions: { type: "integer", minimum: 2 } }, + additionalProperties: false, + }, + ], + messages: { + tooMany: "Boolean expression combines {{count}} conditions; extract it into a named variable.", + }, + }, + create(context) { + const minConditions = context.options[0]?.minConditions ?? DEFAULT_MIN_CONDITIONS; + return { + LogicalExpression(node) { + if (!isBooleanLogical(node)) return; + if (isBooleanLogical(node.parent)) return; + const count = countConditions(node); + if (count < minConditions) return; + context.report({ node, messageId: "tooMany", data: { count } }); + }, + }; + }, +}; + +export default rule; diff --git a/ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/AccessGroupsDetailsPage.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/AccessGroupsDetailsPage.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/AccessGroupsDetailsPage.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/AccessGroupsDetailsPage.test.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/AccessGroupsDetailsPage.tsx b/ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/AccessGroupsDetailsPage.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/AccessGroupsDetailsPage.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/AccessGroupsDetailsPage.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/AccessGroupsModal/AccessGroupBaseForm.tsx b/ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/AccessGroupsModal/AccessGroupBaseForm.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/AccessGroupsModal/AccessGroupBaseForm.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/AccessGroupsModal/AccessGroupBaseForm.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/AccessGroupsModal/AccessGroupCreateModal.tsx b/ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/AccessGroupsModal/AccessGroupCreateModal.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/AccessGroupsModal/AccessGroupCreateModal.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/AccessGroupsModal/AccessGroupCreateModal.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/AccessGroupsModal/AccessGroupEditModal.tsx b/ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/AccessGroupsModal/AccessGroupEditModal.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/AccessGroupsModal/AccessGroupEditModal.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/AccessGroupsModal/AccessGroupEditModal.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/AccessGroupsPage.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/AccessGroupsPage.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/AccessGroupsPage.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/AccessGroupsPage.test.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/AccessGroupsPage.tsx b/ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/AccessGroupsPage.tsx similarity index 94% rename from ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/AccessGroupsPage.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/AccessGroupsPage.tsx index c3dd1d54b32..dbbf4e35900 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/AccessGroupsPage.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/AccessGroupsPage.tsx @@ -19,6 +19,7 @@ import { SortState, TableHeaderSortDropdown, } from "@/components/common_components/TableHeaderSortDropdown/TableHeaderSortDropdown"; +import { DateCell, IdCell } from "@/components/shared/table_cells"; import { AccessGroupDetail } from "./AccessGroupsDetailsPage"; import { AccessGroupCreateModal } from "./AccessGroupsModal/AccessGroupCreateModal"; import { AccessGroup } from "./types"; @@ -143,21 +144,7 @@ export function AccessGroupsPage() { header: () => ID, enableSorting: false, size: 170, - cell: ({ row }) => { - const record = row.original; - return ( - - setSelectedGroupId(record.id)} - > - {record.id} - - - ); - }, + cell: ({ row }) => , }, { id: "name", @@ -211,7 +198,7 @@ export function AccessGroupsPage() { header: () => Created, enableSorting: true, sortingFn: "datetime", - cell: ({ getValue }) => new Date(getValue() as string).toLocaleDateString(), + cell: ({ getValue }) => , meta: { responsive: ["lg"] }, }, { @@ -219,7 +206,7 @@ export function AccessGroupsPage() { accessorKey: "updatedAt", header: () => Updated, enableSorting: false, - cell: ({ getValue }) => new Date(getValue() as string).toLocaleDateString(), + cell: ({ getValue }) => , meta: { responsive: ["xl"] }, }, ...(canModify diff --git a/ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/types.ts b/ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/types.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/access-groups/components/types.ts rename to ui/litellm-dashboard/src/app/(dashboard)/access-groups/_components/types.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/access-groups/page.tsx b/ui/litellm-dashboard/src/app/(dashboard)/access-groups/page.tsx index ae4712b826e..4e9f7031c6d 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/access-groups/page.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/access-groups/page.tsx @@ -1,6 +1,6 @@ "use client"; -import { AccessGroupsPage } from "./components/AccessGroupsPage"; +import { AccessGroupsPage } from "./_components/AccessGroupsPage"; import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; export default function AccessGroups() { diff --git a/ui/litellm-dashboard/src/components/AdminPanel.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/admin-panel/_components/AdminPanel.test.tsx similarity index 96% rename from ui/litellm-dashboard/src/components/AdminPanel.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/admin-panel/_components/AdminPanel.test.tsx index 7d1d2f46cf1..220db23338e 100644 --- a/ui/litellm-dashboard/src/components/AdminPanel.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/admin-panel/_components/AdminPanel.test.tsx @@ -8,34 +8,34 @@ const mockGetAllowedIPs = vi.fn(); const mockAddAllowedIP = vi.fn(); const mockDeleteAllowedIP = vi.fn(); -vi.mock("./networking", () => ({ +vi.mock("@/components/networking", () => ({ getSSOSettings: (...args: unknown[]) => mockGetSSOSettings(...args), getAllowedIPs: (...args: unknown[]) => mockGetAllowedIPs(...args), addAllowedIP: (...args: unknown[]) => mockAddAllowedIP(...args), deleteAllowedIP: (...args: unknown[]) => mockDeleteAllowedIP(...args), })); -vi.mock("./constants", () => ({ +vi.mock("@/components/constants", () => ({ useBaseUrl: () => "http://localhost:4000", })); -vi.mock("./Settings/AdminSettings/SSOSettings/SSOSettings", () => ({ +vi.mock("@/components/Settings/AdminSettings/SSOSettings/SSOSettings", () => ({ default: () =>
SSO Settings
, })); -vi.mock("./Settings/AdminSettings/UISettings/UISettings", () => ({ +vi.mock("@/components/Settings/AdminSettings/UISettings/UISettings", () => ({ default: () =>
UI Settings
, })); -vi.mock("./SCIM", () => ({ +vi.mock("@/components/SCIM", () => ({ default: () =>
SCIM Config
, })); -vi.mock("./SSOModals", () => ({ +vi.mock("@/components/SSOModals", () => ({ default: () =>
SSO Modals
, })); -vi.mock("./UIAccessControlForm", () => ({ +vi.mock("@/components/UIAccessControlForm", () => ({ default: () =>
UI Access Control Form
, })); diff --git a/ui/litellm-dashboard/src/components/AdminPanel.tsx b/ui/litellm-dashboard/src/app/(dashboard)/admin-panel/_components/AdminPanel.tsx similarity index 93% rename from ui/litellm-dashboard/src/components/AdminPanel.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/admin-panel/_components/AdminPanel.tsx index 7867c184ed2..611efd6a588 100644 --- a/ui/litellm-dashboard/src/components/AdminPanel.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/admin-panel/_components/AdminPanel.tsx @@ -16,18 +16,18 @@ import { } from "@tremor/react"; import { Alert, Button as Button2, Form, Input, Modal, Space, Tabs, Typography } from "antd"; import React, { useEffect, useState } from "react"; -import NewBadge from "./common_components/NewBadge"; -import { useBaseUrl } from "./constants"; -import NotificationsManager from "./molecules/notifications_manager"; -import { addAllowedIP, deleteAllowedIP, getAllowedIPs, getSSOSettings } from "./networking"; -import SCIMConfig from "./SCIM"; -import LoggingSettings from "./Settings/AdminSettings/LoggingSettings/LoggingSettings"; -import SSOSettings from "./Settings/AdminSettings/SSOSettings/SSOSettings"; -import UISettings from "./Settings/AdminSettings/UISettings/UISettings"; -import HashicorpVault from "./Settings/AdminSettings/HashicorpVault/HashicorpVault"; -import PluginSettings from "./Settings/AdminSettings/PluginSettings/PluginSettings"; -import SSOModals from "./SSOModals"; -import UIAccessControlForm from "./UIAccessControlForm"; +import NewBadge from "@/components/common_components/NewBadge"; +import { useBaseUrl } from "@/components/constants"; +import NotificationsManager from "@/components/molecules/notifications_manager"; +import { addAllowedIP, deleteAllowedIP, getAllowedIPs, getSSOSettings } from "@/components/networking"; +import SCIMConfig from "@/components/SCIM"; +import LoggingSettings from "@/components/Settings/AdminSettings/LoggingSettings/LoggingSettings"; +import SSOSettings from "@/components/Settings/AdminSettings/SSOSettings/SSOSettings"; +import UISettings from "@/components/Settings/AdminSettings/UISettings/UISettings"; +import HashicorpVault from "@/components/Settings/AdminSettings/HashicorpVault/HashicorpVault"; +import PluginSettings from "@/components/Settings/AdminSettings/PluginSettings/PluginSettings"; +import SSOModals from "@/components/SSOModals"; +import UIAccessControlForm from "@/components/UIAccessControlForm"; const { Title, Paragraph, Text } = Typography; diff --git a/ui/litellm-dashboard/src/app/(dashboard)/admin-panel/page.tsx b/ui/litellm-dashboard/src/app/(dashboard)/admin-panel/page.tsx index aac835b02fc..47076acc9f0 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/admin-panel/page.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/admin-panel/page.tsx @@ -1,6 +1,6 @@ "use client"; -import AdminPanel from "@/components/AdminPanel"; +import AdminPanel from "./_components/AdminPanel"; import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; import useProxySettings from "@/app/(dashboard)/hooks/proxySettings/useProxySettings"; diff --git a/ui/litellm-dashboard/src/app/(dashboard)/api-reference/APIReferenceView.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/api-reference/_components/APIReferenceView.test.tsx similarity index 97% rename from ui/litellm-dashboard/src/app/(dashboard)/api-reference/APIReferenceView.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/api-reference/_components/APIReferenceView.test.tsx index a73973bd742..66fa0dfa63f 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/api-reference/APIReferenceView.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/api-reference/_components/APIReferenceView.test.tsx @@ -2,7 +2,7 @@ import { render } from "@testing-library/react"; import { describe, expect, it, vi } from "vitest"; import APIReferenceView from "./APIReferenceView"; -vi.mock("./components/CodeBlock", () => ({ +vi.mock("@/components/CodeBlock", () => ({ __esModule: true, default: ({ code }: { code: string }) => 
{code}
, })); diff --git a/ui/litellm-dashboard/src/app/(dashboard)/api-reference/APIReferenceView.tsx b/ui/litellm-dashboard/src/app/(dashboard)/api-reference/_components/APIReferenceView.tsx similarity index 97% rename from ui/litellm-dashboard/src/app/(dashboard)/api-reference/APIReferenceView.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/api-reference/_components/APIReferenceView.tsx index 5861cc87e5b..333bd1cad13 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/api-reference/APIReferenceView.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/api-reference/_components/APIReferenceView.tsx @@ -1,8 +1,8 @@ "use client"; import React from "react"; import { Text, Tab, TabGroup, TabList, TabPanel, TabPanels, Grid } from "@tremor/react"; -import CodeBlock from "./components/CodeBlock"; -import DocLink from "@/app/(dashboard)/api-reference/components/DocLink"; +import CodeBlock from "@/components/CodeBlock"; +import DocLink from "./DocLink"; interface ApiRefProps { proxySettings: { diff --git a/ui/litellm-dashboard/src/app/(dashboard)/api-reference/components/DocLink.tsx b/ui/litellm-dashboard/src/app/(dashboard)/api-reference/_components/DocLink.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/api-reference/components/DocLink.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/api-reference/_components/DocLink.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/api-reference/page.tsx b/ui/litellm-dashboard/src/app/(dashboard)/api-reference/page.tsx index 42cf094f0bb..d7c977b0870 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/api-reference/page.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/api-reference/page.tsx @@ -1,6 +1,6 @@ "use client"; -import APIReferenceView from "@/app/(dashboard)/api-reference/APIReferenceView"; +import APIReferenceView from "./_components/APIReferenceView"; import { DeprecationBanner } from "@/components/DeprecationBanner"; import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; import useProxySettings from "@/app/(dashboard)/hooks/proxySettings/useProxySettings"; diff --git a/ui/litellm-dashboard/src/app/(dashboard)/budgets/components/budget_modal.tsx b/ui/litellm-dashboard/src/app/(dashboard)/budgets/_components/budget_modal.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/budgets/components/budget_modal.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/budgets/_components/budget_modal.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/budgets/components/budget_panel.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/budgets/_components/budget_panel.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/budgets/components/budget_panel.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/budgets/_components/budget_panel.test.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/budgets/components/budget_panel.tsx b/ui/litellm-dashboard/src/app/(dashboard)/budgets/_components/budget_panel.tsx similarity index 96% rename from ui/litellm-dashboard/src/app/(dashboard)/budgets/components/budget_panel.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/budgets/_components/budget_panel.tsx index 0e601645c20..af15a99f0b4 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/budgets/components/budget_panel.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/budgets/_components/budget_panel.tsx @@ -25,6 +25,7 @@ import DeleteResourceModal from "@/components/common_components/DeleteResourceMo import TableIconActionButton from "@/components/common_components/IconActionButton/TableIconActionButtons/TableIconActionButton"; import NotificationsManager from "@/components/molecules/notifications_manager"; import { useBudgets, useDeleteBudget, budgetItem } from "@/app/(dashboard)/hooks/budgets/useBudgets"; +import { MoneyCell } from "@/components/shared/table_cells"; import BudgetModal from "./budget_modal"; import EditBudgetModal from "./edit_budget_modal"; import { CREATE_END_USER_CURL_COMMAND, CHAT_COMPLETIONS_CURL_COMMAND, OPENAI_SDK_PYTHON_CODE } from "./constants"; @@ -127,7 +128,9 @@ const BudgetPanel: React.FC = ({ accessToken }) => { .map((value: budgetItem) => ( {value.budget_id} - {value.max_budget ? value.max_budget : "n/a"} + + + {value.tpm_limit ? value.tpm_limit : "n/a"} {value.rpm_limit ? value.rpm_limit : "n/a"} {canModify && ( diff --git a/ui/litellm-dashboard/src/app/(dashboard)/budgets/components/constants.ts b/ui/litellm-dashboard/src/app/(dashboard)/budgets/_components/constants.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/budgets/components/constants.ts rename to ui/litellm-dashboard/src/app/(dashboard)/budgets/_components/constants.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/budgets/components/edit_budget_modal.tsx b/ui/litellm-dashboard/src/app/(dashboard)/budgets/_components/edit_budget_modal.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/budgets/components/edit_budget_modal.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/budgets/_components/edit_budget_modal.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/budgets/page.tsx b/ui/litellm-dashboard/src/app/(dashboard)/budgets/page.tsx index 547699411e7..ca34589a679 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/budgets/page.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/budgets/page.tsx @@ -1,6 +1,6 @@ "use client"; -import BudgetPanel from "./components/budget_panel"; +import BudgetPanel from "./_components/budget_panel"; import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; export default function Budgets() { diff --git a/ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_dashboard.tsx b/ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_dashboard.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_dashboard.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_dashboard.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_health.tsx b/ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_health.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_health.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_health.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/CacheFieldSection.tsx b/ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/CacheFieldSection.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/CacheFieldSection.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/CacheFieldSection.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/CacheFormField.tsx b/ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/CacheFormField.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/CacheFormField.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/CacheFormField.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/RedisTypeSelector.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/RedisTypeSelector.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/RedisTypeSelector.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/RedisTypeSelector.test.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/RedisTypeSelector.tsx b/ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/RedisTypeSelector.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/RedisTypeSelector.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/RedisTypeSelector.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/cacheSettingsFields.ts b/ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/cacheSettingsFields.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/cacheSettingsFields.ts rename to ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/cacheSettingsFields.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/cacheSettingsUtils.test.ts b/ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/cacheSettingsUtils.test.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/cacheSettingsUtils.test.ts rename to ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/cacheSettingsUtils.test.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/cacheSettingsUtils.ts b/ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/cacheSettingsUtils.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/cacheSettingsUtils.ts rename to ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/cacheSettingsUtils.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/index.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/index.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/index.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/index.test.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/index.tsx b/ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/index.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/caching/components/cache_settings/index.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/caching/_components/cache_settings/index.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/caching/components/response_time_indicator.tsx b/ui/litellm-dashboard/src/app/(dashboard)/caching/_components/response_time_indicator.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/caching/components/response_time_indicator.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/caching/_components/response_time_indicator.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/caching/page.tsx b/ui/litellm-dashboard/src/app/(dashboard)/caching/page.tsx index 0ef88ec9eb5..33f3e81c689 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/caching/page.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/caching/page.tsx @@ -1,6 +1,6 @@ "use client"; -import CacheDashboard from "./components/cache_dashboard"; +import CacheDashboard from "./_components/cache_dashboard"; import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; export default function Caching() { diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/add_margin_form.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/add_margin_form.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/add_margin_form.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/add_margin_form.test.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/add_margin_form.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/add_margin_form.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/add_margin_form.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/add_margin_form.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/add_provider_form.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/add_provider_form.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/add_provider_form.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/add_provider_form.test.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/add_provider_form.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/add_provider_form.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/add_provider_form.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/add_provider_form.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/cost_tracking_settings.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/cost_tracking_settings.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/cost_tracking_settings.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/cost_tracking_settings.test.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/cost_tracking_settings.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/cost_tracking_settings.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/cost_tracking_settings.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/cost_tracking_settings.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/how_it_works.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/how_it_works.test.tsx similarity index 98% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/how_it_works.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/how_it_works.test.tsx index 711a8795f15..a574f4b628e 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/how_it_works.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/how_it_works.test.tsx @@ -5,7 +5,7 @@ import userEvent from "@testing-library/user-event"; import { renderWithProviders } from "../../../../../tests/test-utils"; import HowItWorks from "./how_it_works"; -vi.mock("@/app/(dashboard)/api-reference/components/CodeBlock", () => ({ +vi.mock("@/components/CodeBlock", () => ({ default: ({ code }: { code: string }) => 
{code}
, })); diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/how_it_works.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/how_it_works.tsx similarity index 98% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/how_it_works.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/how_it_works.tsx index 79abf6baa31..5fa27551d16 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/how_it_works.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/how_it_works.tsx @@ -1,6 +1,6 @@ import React, { useState, useMemo } from "react"; import { Text, TextInput } from "@tremor/react"; -import CodeBlock from "@/app/(dashboard)/api-reference/components/CodeBlock"; +import CodeBlock from "@/components/CodeBlock"; const HowItWorks: React.FC = () => { const [responseCost, setResponseCost] = useState(""); diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/index.ts b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/index.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/index.ts rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/index.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/index.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/index.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/index.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/index.test.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/index.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/index.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/index.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/index.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_cost_results.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_cost_results.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_cost_results.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_cost_results.test.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_cost_results.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_cost_results.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_cost_results.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_cost_results.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_export_dropdown.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_export_dropdown.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_export_dropdown.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_export_dropdown.test.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_export_dropdown.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_export_dropdown.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_export_dropdown.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_export_dropdown.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_export_utils.test.ts b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_export_utils.test.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_export_utils.test.ts rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_export_utils.test.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_export_utils.ts b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_export_utils.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/multi_export_utils.ts rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/multi_export_utils.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/types.ts b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/types.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/types.ts rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/types.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/use_multi_cost_estimate.test.ts b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/use_multi_cost_estimate.test.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/use_multi_cost_estimate.test.ts rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/use_multi_cost_estimate.test.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/use_multi_cost_estimate.ts b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/use_multi_cost_estimate.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/pricing_calculator/use_multi_cost_estimate.ts rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/pricing_calculator/use_multi_cost_estimate.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/provider_discount_table.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/provider_discount_table.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/provider_discount_table.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/provider_discount_table.test.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/provider_discount_table.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/provider_discount_table.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/provider_discount_table.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/provider_discount_table.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/provider_display_helpers.test.ts b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/provider_display_helpers.test.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/provider_display_helpers.test.ts rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/provider_display_helpers.test.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/provider_display_helpers.ts b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/provider_display_helpers.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/provider_display_helpers.ts rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/provider_display_helpers.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/provider_margin_table.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/provider_margin_table.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/provider_margin_table.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/provider_margin_table.test.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/provider_margin_table.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/provider_margin_table.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/provider_margin_table.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/provider_margin_table.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/types.ts b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/types.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/types.ts rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/types.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/use_discount_config.test.ts b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/use_discount_config.test.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/use_discount_config.test.ts rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/use_discount_config.test.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/use_discount_config.ts b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/use_discount_config.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/use_discount_config.ts rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/use_discount_config.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/use_margin_config.test.ts b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/use_margin_config.test.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/use_margin_config.test.ts rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/use_margin_config.test.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/use_margin_config.ts b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/use_margin_config.ts similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/components/use_margin_config.ts rename to ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/_components/use_margin_config.ts diff --git a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/page.tsx b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/page.tsx index c72fed4c594..6c251ca61d0 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/page.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/cost-tracking/page.tsx @@ -1,6 +1,6 @@ "use client"; -import { CostTrackingSettings } from "./components"; +import { CostTrackingSettings } from "./_components"; import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; export default function CostTracking() { diff --git a/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/EvaluationSettingsModal.tsx b/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/EvaluationSettingsModal.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/EvaluationSettingsModal.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/EvaluationSettingsModal.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/GuardrailConfig.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/GuardrailConfig.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/GuardrailConfig.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/GuardrailConfig.test.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/GuardrailConfig.tsx b/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/GuardrailConfig.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/GuardrailConfig.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/GuardrailConfig.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/GuardrailDetail.tsx b/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/GuardrailDetail.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/GuardrailDetail.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/GuardrailDetail.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/GuardrailsMonitorView.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/GuardrailsMonitorView.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/GuardrailsMonitorView.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/GuardrailsMonitorView.test.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/GuardrailsMonitorView.tsx b/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/GuardrailsMonitorView.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/GuardrailsMonitorView.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/GuardrailsMonitorView.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/GuardrailsOverview.tsx b/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/GuardrailsOverview.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/GuardrailsOverview.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/GuardrailsOverview.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/ScoreChart.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/ScoreChart.test.tsx new file mode 100644 index 00000000000..dba34ea9a86 --- /dev/null +++ b/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/ScoreChart.test.tsx @@ -0,0 +1,42 @@ +import React from "react"; +import { describe, it, expect } from "vitest"; +import { screen } from "@testing-library/react"; +import { renderWithProviders } from "../../../../../tests/test-utils"; +import { ScoreChart } from "./ScoreChart"; + +describe("ScoreChart", () => { + it("should render the title", () => { + renderWithProviders(); + + expect(screen.getByText("Request Outcomes Over Time")).toBeInTheDocument(); + }); + + it("should show empty state when no data is provided", () => { + renderWithProviders(); + + expect(screen.getByText("No chart data for this period")).toBeInTheDocument(); + }); + + it("should show empty state when data is an empty array", () => { + renderWithProviders(); + + expect(screen.getByText("No chart data for this period")).toBeInTheDocument(); + }); + + it("should render the chart when data is provided", () => { + const data = [ + { date: "2026-03-01", passed: 10, blocked: 2 }, + { date: "2026-03-02", passed: 15, blocked: 1 }, + ]; + + const { container } = renderWithProviders(); + + expect(screen.queryByText("No chart data for this period")).not.toBeInTheDocument(); + expect(screen.getByText("passed")).toBeInTheDocument(); + expect(screen.getByText("blocked")).toBeInTheDocument(); + expect(screen.getAllByText(/2026-03-01/).length).toBeGreaterThan(0); + expect(screen.getAllByText(/2026-03-02/).length).toBeGreaterThan(0); + const bars = container.querySelectorAll(".recharts-bar"); + expect(bars).toHaveLength(2); + }); +}); diff --git a/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/ScoreChart.tsx b/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/ScoreChart.tsx new file mode 100644 index 00000000000..bc11a6fd3e0 --- /dev/null +++ b/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/_components/ScoreChart.tsx @@ -0,0 +1,44 @@ +import React from "react"; +import { BarChart } from "@/components/shared/charts"; +import { Card, CardContent, CardHeader, CardTitle } from "@/components/ui/card"; + +/** + * Overview chart: Request Outcomes Over Time (passed vs blocked). + * Stacked bar chart. Data from usage/overview API (chart array). + */ +interface ScoreChartProps { + data?: Array<{ date: string; passed: number; blocked: number }>; +} + +export function ScoreChart({ data }: ScoreChartProps) { + const chartData = data && data.length > 0 ? data : []; + + return ( + + + Request Outcomes Over Time + + +
+ {chartData.length > 0 ? ( + v.toLocaleString()} + yAxisWidth={48} + showLegend={true} + stack={true} + className="h-full" + /> + ) : ( +
+ No chart data for this period +
+ )} +
+ + + ); +} diff --git a/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/ScoreChart.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/ScoreChart.test.tsx deleted file mode 100644 index 3a36eb9621e..00000000000 --- a/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/ScoreChart.test.tsx +++ /dev/null @@ -1,64 +0,0 @@ -import React from "react"; -import { describe, it, expect, vi } from "vitest"; -import { screen } from "@testing-library/react"; -import { renderWithProviders } from "../../../../../tests/test-utils"; -import { ScoreChart } from "./ScoreChart"; - -vi.mock("@tremor/react", async (importOriginal) => { - const actual = await importOriginal(); - // Re-apply the global Button/Tooltip overrides from tests/setupTests.ts. A file-level - // vi.mock fully replaces the setup-level mock, so without this the real Tremor Button - // leaks through and its useTooltip(300) schedules a native setTimeout that can fire - // post-teardown -> "window is not defined". - return { - ...actual, - BarChart: ({ data, categories }: { data: any[]; categories: string[] }) => ( -
- {data.map((d, i) => ( - - {d.date}: {categories.map((c) => `${c}=${d[c]}`).join(", ")} - - ))} -
- ), - Button: React.forwardRef(({ children, ...props }, ref) => ( - - )), - Tooltip: ({ children }: { children?: React.ReactNode }) => <>{children}, - }; -}); - -describe("ScoreChart", () => { - it("should render the title", () => { - renderWithProviders(); - - expect(screen.getByText("Request Outcomes Over Time")).toBeInTheDocument(); - }); - - it("should show empty state when no data is provided", () => { - renderWithProviders(); - - expect(screen.getByText("No chart data for this period")).toBeInTheDocument(); - }); - - it("should show empty state when data is an empty array", () => { - renderWithProviders(); - - expect(screen.getByText("No chart data for this period")).toBeInTheDocument(); - }); - - it("should render the chart when data is provided", () => { - const data = [ - { date: "2026-03-01", passed: 10, blocked: 2 }, - { date: "2026-03-02", passed: 15, blocked: 1 }, - ]; - - renderWithProviders(); - - expect(screen.queryByText("No chart data for this period")).not.toBeInTheDocument(); - expect(screen.getByText(/2026-03-01/)).toBeInTheDocument(); - expect(screen.getByText(/2026-03-02/)).toBeInTheDocument(); - }); -}); diff --git a/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/ScoreChart.tsx b/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/ScoreChart.tsx deleted file mode 100644 index daa6054a552..00000000000 --- a/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/components/ScoreChart.tsx +++ /dev/null @@ -1,38 +0,0 @@ -import { BarChart, Card, Title } from "@tremor/react"; -import React from "react"; - -/** - * Overview chart: Request Outcomes Over Time (passed vs blocked). - * Uses Tremor BarChart with stacked data. Data from usage/overview API (chart array). - */ -interface ScoreChartProps { - data?: Array<{ date: string; passed: number; blocked: number }>; -} - -export function ScoreChart({ data }: ScoreChartProps) { - const chartData = data && data.length > 0 ? data : []; - - return ( - - Request Outcomes Over Time -
- {chartData.length > 0 ? ( - v.toLocaleString()} - yAxisWidth={48} - showLegend={true} - stack={true} - /> - ) : ( -
- No chart data for this period -
- )} -
- - ); -} diff --git a/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/page.tsx b/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/page.tsx index 388ed168f17..0c4e69c2d80 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/page.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/guardrails-monitor/page.tsx @@ -1,6 +1,6 @@ "use client"; -import GuardrailsMonitorView from "./components/GuardrailsMonitorView"; +import GuardrailsMonitorView from "./_components/GuardrailsMonitorView"; import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; export default function GuardrailsMonitor() { diff --git a/ui/litellm-dashboard/src/app/(dashboard)/hooks/license/useLicenseInfo.ts b/ui/litellm-dashboard/src/app/(dashboard)/hooks/license/useLicenseInfo.ts new file mode 100644 index 00000000000..3ea0bd20e40 --- /dev/null +++ b/ui/litellm-dashboard/src/app/(dashboard)/hooks/license/useLicenseInfo.ts @@ -0,0 +1,16 @@ +import { useQuery, UseQueryResult } from "@tanstack/react-query"; +import { getLicenseInfo, LicenseInfo } from "@/components/networking"; +import { createQueryKeys } from "../common/queryKeysFactory"; + +const licenseInfoKeys = createQueryKeys("licenseInfo"); + +export const useLicenseInfo = (accessToken: string | null | undefined): UseQueryResult => { + const options = { + queryKey: licenseInfoKeys.detail("license"), + queryFn: () => getLicenseInfo(accessToken!), + enabled: Boolean(accessToken), + staleTime: 5 * 60 * 1000, + retry: false, + }; + return useQuery(options); +}; diff --git a/ui/litellm-dashboard/src/app/(dashboard)/layout.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/layout.test.tsx index af68d9f87e9..92a1d40b0e3 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/layout.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/layout.test.tsx @@ -3,9 +3,13 @@ import { render, screen, waitFor } from "@testing-library/react"; import { AuthProvider } from "@/contexts/AuthContext"; import Layout from "./layout"; +const { replaceMock } = vi.hoisted(() => ({ replaceMock: vi.fn() })); + +let searchParamsValue = new URLSearchParams(); + vi.mock("next/navigation", () => ({ - useRouter: vi.fn(() => ({ push: vi.fn(), replace: vi.fn() })), - useSearchParams: vi.fn(() => new URLSearchParams()), + useRouter: vi.fn(() => ({ push: vi.fn(), replace: replaceMock })), + useSearchParams: vi.fn(() => searchParamsValue), usePathname: vi.fn(() => "/ui/guardrails"), })); @@ -21,6 +25,10 @@ vi.mock("@/components/DebugWarningBanner", () => ({ DebugWarningBanner: () => null, })); +vi.mock("@/components/LicenseExpiryBanner", () => ({ + LicenseExpiryBanner: () => null, +})); + vi.mock("@/contexts/ThemeContext", () => ({ ThemeProvider: ({ children }: { children: React.ReactNode }) => <>{children}, })); @@ -54,6 +62,7 @@ describe("(dashboard) Layout", () => { beforeEach(() => { vi.clearAllMocks(); pendingUiConfig = createDeferred(); + searchParamsValue = new URLSearchParams(); }); it("does not mount route content until getUiConfig has resolved", async () => { @@ -75,4 +84,25 @@ describe("(dashboard) Layout", () => { expect(screen.getByTestId("navbar")).toBeTruthy(); expect(screen.queryByTestId("loading-screen")).toBeNull(); }); + + it("redirects an invitation link to the onboarding route instead of rendering the dashboard shell", async () => { + searchParamsValue = new URLSearchParams("invitation_id=abc123"); + + render( + + +
+ + , + ); + + pendingUiConfig.resolve(); + + await waitFor(() => + expect(replaceMock).toHaveBeenCalledWith(expect.stringContaining("/onboarding?invitation_id=abc123")), + ); + expect(screen.queryByTestId("page-content")).toBeNull(); + expect(screen.queryByTestId("navbar")).toBeNull(); + expect(screen.queryByTestId("sidebar")).toBeNull(); + }); }); diff --git a/ui/litellm-dashboard/src/app/(dashboard)/layout.tsx b/ui/litellm-dashboard/src/app/(dashboard)/layout.tsx index 09951dc1923..b8eb4e66ed0 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/layout.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/layout.tsx @@ -8,6 +8,7 @@ import { useAuth } from "@/contexts/AuthContext"; import SidebarProvider from "@/app/(dashboard)/components/SidebarProvider"; import { useRouter, useSearchParams, usePathname } from "next/navigation"; import { DebugWarningBanner } from "@/components/DebugWarningBanner"; +import { LicenseExpiryBanner } from "@/components/LicenseExpiryBanner"; import { MIGRATED_PAGES, migratedHref, legacyPageHref, legacyKeyForPathname } from "@/utils/migratedPages"; import { PluginModeProvider, usePluginMode } from "@/contexts/PluginModeContext"; import { createApiClient } from "@/lib/http/client"; @@ -116,6 +117,7 @@ function DashboardShell({ children }: { children: React.ReactNode }) { onToggleSidebar={() => setSidebarCollapsed((v) => !v)} /> +
{mode !== "ai-gateway" ? (
@@ -135,17 +137,26 @@ function DashboardShell({ children }: { children: React.ReactNode }) { } function LayoutContent({ children }: { children: React.ReactNode }) { + const router = useRouter(); const searchParams = useSearchParams(); const { accessToken, authLoading } = useAuth(); const isInvitationFlow = Boolean(searchParams.get("invitation_id")); - if (authLoading) { + // Legacy invitation links point at /ui/?invitation_id=; the onboarding form now lives at its own + // /onboarding route. Redirect once ui-config has loaded so migratedHref resolves the SERVER_ROOT_PATH base. + useEffect(() => { + if (!authLoading && isInvitationFlow) { + router.replace(`${migratedHref("onboarding")}?${searchParams.toString()}`); + } + }, [authLoading, isInvitationFlow, router, searchParams]); + + if (authLoading || isInvitationFlow) { return ; } return ( - {isInvitationFlow ? children : {children}} + {children} ); } diff --git a/ui/litellm-dashboard/src/app/(dashboard)/memory/components/MemoryEditModal.tsx b/ui/litellm-dashboard/src/app/(dashboard)/memory/_components/MemoryEditModal.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/memory/components/MemoryEditModal.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/memory/_components/MemoryEditModal.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/memory/components/MemoryView.tsx b/ui/litellm-dashboard/src/app/(dashboard)/memory/_components/MemoryView.tsx similarity index 92% rename from ui/litellm-dashboard/src/app/(dashboard)/memory/components/MemoryView.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/memory/_components/MemoryView.tsx index 402de29e0c5..4ee784f4664 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/memory/components/MemoryView.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/memory/_components/MemoryView.tsx @@ -2,7 +2,7 @@ import React, { useMemo, useState } from "react"; import { useMutation, useQuery, useQueryClient } from "@tanstack/react-query"; -import { Button, Card, Drawer, Empty, Input, Space, Table, Tooltip, Typography, message } from "antd"; +import { Button, Card, Drawer, Empty, Input, Space, Table, Typography, message } from "antd"; import type { ColumnsType } from "antd/es/table"; import { DeleteOutlined, @@ -13,6 +13,7 @@ import { SearchOutlined, } from "@ant-design/icons"; import { MemoryRow, createMemory, deleteMemory, fetchMemoryList, updateMemory } from "@/components/networking"; +import { DateCell, IdCell } from "@/components/shared/table_cells"; import { MemoryEditModal } from "./MemoryEditModal"; import DeleteResourceModal from "@/components/common_components/DeleteResourceModal"; @@ -191,34 +192,13 @@ export const MemoryView: React.FC = ({ accessToken }) => { } }; - const renderIdPill = (id: string | null | undefined, onClick?: () => void) => { - if (!id) return -; - const short = id.length > 10 ? `${id.slice(0, 7)}...` : id; - const pillClass = - "font-mono text-blue-600 bg-blue-50 text-xs font-medium px-2 py-0.5 rounded-md border border-blue-200 inline-block max-w-[15ch] truncate whitespace-nowrap"; - return ( - - {onClick ? ( - - ) : ( - {short} - )} - - ); - }; - const columns: ColumnsType = [ { title: "ID", dataIndex: "memory_id", key: "memory_id", width: 140, - render: (_: unknown, r: MemoryRow) => renderIdPill(r.memory_id, () => setDetailRow(r)), + render: (_: unknown, r: MemoryRow) => setDetailRow(r)} />, }, { title: "Name", @@ -246,21 +226,21 @@ export const MemoryView: React.FC = ({ accessToken }) => { dataIndex: "user_id", key: "user_id", width: 160, - render: (uid?: string | null) => renderIdPill(uid), + render: (uid?: string | null) => , }, { title: "Team ID", dataIndex: "team_id", key: "team_id", width: 160, - render: (tid?: string | null) => renderIdPill(tid), + render: (tid?: string | null) => , }, { title: "Updated", dataIndex: "updated_at", key: "updated_at", width: 180, - render: (ts?: string) => {formatTimestamp(ts)}, + render: (ts?: string) => , // No sorter — backend already returns rows in `updated_at DESC` order, // and a client-side sorter on a paginated view would only affect the // current page. diff --git a/ui/litellm-dashboard/src/app/(dashboard)/memory/page.tsx b/ui/litellm-dashboard/src/app/(dashboard)/memory/page.tsx index 031a027d518..b88996c5396 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/memory/page.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/memory/page.tsx @@ -1,6 +1,6 @@ "use client"; -import { MemoryView } from "./components/MemoryView"; +import { MemoryView } from "./_components/MemoryView"; import { DeprecationBanner } from "@/components/DeprecationBanner"; import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; diff --git a/ui/litellm-dashboard/src/components/usage.tsx b/ui/litellm-dashboard/src/app/(dashboard)/old-usage/_components/usage.tsx similarity index 98% rename from ui/litellm-dashboard/src/components/usage.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/old-usage/_components/usage.tsx index 04fcbddbd37..01f8cb1cd45 100644 --- a/ui/litellm-dashboard/src/components/usage.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/old-usage/_components/usage.tsx @@ -14,9 +14,9 @@ import { import React, { useState, useEffect } from "react"; -import ViewUserSpend from "./view_user_spend"; -import { ProxySettings } from "./user_dashboard"; -import UsageDatePicker from "./shared/usage_date_picker"; +import ViewUserSpend from "@/components/view_user_spend"; +import { ProxySettings } from "@/components/user_dashboard"; +import UsageDatePicker from "@/components/shared/usage_date_picker"; import { Grid, Col, @@ -48,8 +48,9 @@ import { adminGlobalActivity, adminGlobalActivityPerModel, getProxyUISettings, -} from "./networking"; -import TopKeyView from "./UsagePage/components/EntityUsage/TopKeyView"; +} from "@/components/networking"; +import TopKeyView from "@/components/UsagePage/components/EntityUsage/TopKeyView"; +import { MoneyCell } from "@/components/shared/table_cells"; import { formatNumberWithCommas } from "@/utils/dataUtils"; interface UsagePageProps { @@ -644,9 +645,7 @@ const UsagePage: React.FC = ({ accessToken, token, userRole, use {provider.provider} - {parseFloat(provider.spend.toFixed(2)) < 0.00001 - ? "less than 0.00" - : formatNumberWithCommas(provider.spend, 2)} + ))} @@ -819,7 +818,9 @@ const UsagePage: React.FC = ({ accessToken, token, userRole, use {topUsers?.map((user: any, index: number) => ( {user.end_user} - {formatNumberWithCommas(user.total_spend, 2)} + + + {user.total_count} ))} diff --git a/ui/litellm-dashboard/src/app/(dashboard)/old-usage/page.tsx b/ui/litellm-dashboard/src/app/(dashboard)/old-usage/page.tsx index cc1f2c35e44..138dd97e5e8 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/old-usage/page.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/old-usage/page.tsx @@ -1,6 +1,6 @@ "use client"; -import Usage from "@/components/usage"; +import Usage from "./_components/usage"; import { DeprecationBanner } from "@/components/DeprecationBanner"; import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; diff --git a/ui/litellm-dashboard/src/components/organizations.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/organizations/_components/organizations.test.tsx similarity index 87% rename from ui/litellm-dashboard/src/components/organizations.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/organizations/_components/organizations.test.tsx index 9be31be6170..75a6d30ac2e 100644 --- a/ui/litellm-dashboard/src/components/organizations.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/organizations/_components/organizations.test.tsx @@ -3,11 +3,11 @@ import { render } from "@testing-library/react"; import React from "react"; import { describe, expect, it, vi } from "vitest"; -vi.mock("./vector_store_management/VectorStoreSelector", () => ({ +vi.mock("@/components/vector_store_management/VectorStoreSelector", () => ({ __esModule: true, default: () => null, })); -vi.mock("./mcp_server_management/MCPServerSelector", () => ({ +vi.mock("@/components/mcp_server_management/MCPServerSelector", () => ({ __esModule: true, default: () => null, })); diff --git a/ui/litellm-dashboard/src/components/organizations.tsx b/ui/litellm-dashboard/src/app/(dashboard)/organizations/_components/organizations.tsx similarity index 90% rename from ui/litellm-dashboard/src/components/organizations.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/organizations/_components/organizations.tsx index 857a0a0c64c..d3af5b62668 100644 --- a/ui/litellm-dashboard/src/components/organizations.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/organizations/_components/organizations.tsx @@ -27,17 +27,22 @@ import { import { Form, Input, Modal, Select as Select2, Tooltip } from "antd"; import { useQueryClient } from "@tanstack/react-query"; import React, { useState } from "react"; -import { formatNumberWithCommas } from "../utils/dataUtils"; -import DeleteResourceModal from "./common_components/DeleteResourceModal"; -import TableIconActionButton from "./common_components/IconActionButton/TableIconActionButtons/TableIconActionButton"; -import { getModelDisplayName } from "./key_team_helpers/fetch_available_models_team_key"; -import MCPServerSelector from "./mcp_server_management/MCPServerSelector"; -import { ModelSelect } from "./ModelSelect/ModelSelect"; -import NotificationsManager from "./molecules/notifications_manager"; -import { Organization, organizationCreateCall, organizationDeleteCall, organizationListCall } from "./networking"; -import OrganizationInfoView from "./organization/organization_view"; -import NumericalInput from "./shared/numerical_input"; -import VectorStoreSelector from "./vector_store_management/VectorStoreSelector"; +import { DateCell, IdCell, MoneyCell } from "@/components/shared/table_cells"; +import DeleteResourceModal from "@/components/common_components/DeleteResourceModal"; +import TableIconActionButton from "@/components/common_components/IconActionButton/TableIconActionButtons/TableIconActionButton"; +import { getModelDisplayName } from "@/components/key_team_helpers/fetch_available_models_team_key"; +import MCPServerSelector from "@/components/mcp_server_management/MCPServerSelector"; +import { ModelSelect } from "@/components/ModelSelect/ModelSelect"; +import NotificationsManager from "@/components/molecules/notifications_manager"; +import { + Organization, + organizationCreateCall, + organizationDeleteCall, + organizationListCall, +} from "@/components/networking"; +import OrganizationInfoView from "@/components/organization/organization_view"; +import NumericalInput from "@/components/shared/numerical_input"; +import VectorStoreSelector from "@/components/vector_store_management/VectorStoreSelector"; interface OrganizationsTableProps { userRole: string; @@ -263,30 +268,22 @@ const OrganizationsTable: React.FC = ({ .map((org: Organization) => ( -
- - - -
+ {org.organization_alias} - {org.created_at ? new Date(org.created_at).toLocaleDateString() : "N/A"} + - {formatNumberWithCommas(org.spend, 4)} - {org.litellm_budget_table?.max_budget !== null && - org.litellm_budget_table?.max_budget !== undefined - ? org.litellm_budget_table?.max_budget - : "No limit"} + + + + (null); - const [keys, setKeys] = useState([]); + const { authLoading, token } = useAuth(); const router = useRouter(); const searchParams = useSearchParams()!; - const [createClicked, setCreateClicked] = useState(false); - - const invitation_id = searchParams.get("invitation_id"); const explicitPage = searchParams.get("page"); - const page = explicitPage || "api-keys"; // Track if we've already attempted a return URL redirect to prevent race conditions const hasAttemptedReturnRedirectRef = useRef(false); - const addKey = (data: any) => { - setKeys((prevData) => (prevData ? [...prevData, data] : [data])); - setCreateClicked(() => !createClicked); - }; - const redirectToLogin = authLoading === false && token === null && invitation_id === null; + const redirectToLogin = authLoading === false && token === null; useEffect(() => { if (redirectToLogin) { @@ -55,15 +40,13 @@ function CreateKeyPageContent() { } }, [redirectToLogin]); - // Redirect legacy query-param pages to their new path-based routes. Only when the page is - // explicitly requested via ?page=, so the bare landing renders inline and the post-login - // return-URL handling below stays intact. + // Redirect legacy ?page= deep links (old bookmarks) to their path-based routes. const isLegacyRedirect = explicitPage !== null && explicitPage in MIGRATED_PAGES; useEffect(() => { if (!authLoading && isLegacyRedirect) { - router.replace(migratedHref(MIGRATED_PAGES[page])); + router.replace(migratedHref(MIGRATED_PAGES[explicitPage])); } - }, [authLoading, isLegacyRedirect, page, router]); + }, [authLoading, isLegacyRedirect, explicitPage, router]); // Check for a stored return URL after successful authentication // This handles the case where user comes back from SSO and we need to redirect to the original URL @@ -102,42 +85,11 @@ function CreateKeyPageContent() { } }, [token]); - useEffect(() => { - if (accessToken && userID && userRole) { - v2TeamListCall(accessToken, 1, 100, { - userID: userRole !== "Admin" && userRole !== "Admin Viewer" ? userID : null, - }) - .then((response) => setTeams(response.teams ?? [])) - .catch(console.error); - } - }, [accessToken, userID, userRole]); - if (authLoading || redirectToLogin || isLegacyRedirect) { return ; } - return ( - <> - {invitation_id ? ( - - ) : ( - - )} - - ); + return ; } export default function CreateKeyPage() { diff --git a/ui/litellm-dashboard/src/app/(dashboard)/playground/components/chat_ui/AgentBuilderView.tsx b/ui/litellm-dashboard/src/app/(dashboard)/playground/components/chat_ui/AgentBuilderView.tsx index 35f5dcf06c0..d4333b95c62 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/playground/components/chat_ui/AgentBuilderView.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/playground/components/chat_ui/AgentBuilderView.tsx @@ -11,7 +11,7 @@ import { } from "@ant-design/icons"; import { Button, Input, Modal, Select, Spin, Tabs } from "antd"; import React, { useCallback, useEffect, useState } from "react"; -import CodeBlock from "@/app/(dashboard)/api-reference/components/CodeBlock"; +import CodeBlock from "@/components/CodeBlock"; import NotificationsManager from "@/components/molecules/notifications_manager"; import { keyCreateCall, diff --git a/ui/litellm-dashboard/src/components/policies/add_attachment_form.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/policies/_components/add_attachment_form.test.tsx similarity index 96% rename from ui/litellm-dashboard/src/components/policies/add_attachment_form.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/policies/_components/add_attachment_form.test.tsx index adbbb1b6997..6788e97d25d 100644 --- a/ui/litellm-dashboard/src/components/policies/add_attachment_form.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/policies/_components/add_attachment_form.test.tsx @@ -1,15 +1,15 @@ import React from "react"; import { screen, waitFor, within } from "@testing-library/react"; import userEvent from "@testing-library/user-event"; -import { renderWithProviders } from "../../../tests/test-utils"; +import { renderWithProviders } from "@/../tests/test-utils"; import { beforeEach, describe, expect, it, vi } from "vitest"; -import * as networking from "../networking"; +import * as networking from "@/components/networking"; import AddAttachmentForm from "./add_attachment_form"; -import { Policy } from "./types"; +import { Policy } from "@/components/policies/types"; -vi.mock("../networking"); +vi.mock("@/components/networking"); -vi.mock("../molecules/notifications_manager", () => ({ +vi.mock("@/components/molecules/notifications_manager", () => ({ default: { success: vi.fn(), fromBackend: vi.fn(), error: vi.fn(), info: vi.fn() }, })); diff --git a/ui/litellm-dashboard/src/components/policies/add_attachment_form.tsx b/ui/litellm-dashboard/src/app/(dashboard)/policies/_components/add_attachment_form.tsx similarity index 98% rename from ui/litellm-dashboard/src/components/policies/add_attachment_form.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/policies/_components/add_attachment_form.tsx index dd4d07b3f35..57203e9ccce 100644 --- a/ui/litellm-dashboard/src/components/policies/add_attachment_form.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/policies/_components/add_attachment_form.tsx @@ -1,9 +1,9 @@ import React, { useState, useEffect } from "react"; import { Modal, Form, Select, Radio, Divider, Typography } from "antd"; import { Button } from "@tremor/react"; -import { Policy } from "./types"; -import { teamListCall, keyListCall, modelAvailableCall, estimateAttachmentImpactCall } from "../networking"; -import NotificationsManager from "../molecules/notifications_manager"; +import { Policy } from "@/components/policies/types"; +import { teamListCall, keyListCall, modelAvailableCall, estimateAttachmentImpactCall } from "@/components/networking"; +import NotificationsManager from "@/components/molecules/notifications_manager"; import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; import { buildAttachmentData } from "./build_attachment_data"; import { getInvalidTeamEntries } from "./scope_validation"; diff --git a/ui/litellm-dashboard/src/components/policies/add_policy_form.tsx b/ui/litellm-dashboard/src/app/(dashboard)/policies/_components/add_policy_form.tsx similarity index 98% rename from ui/litellm-dashboard/src/components/policies/add_policy_form.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/policies/_components/add_policy_form.tsx index 744f77c74a7..8da2d3af36f 100644 --- a/ui/litellm-dashboard/src/components/policies/add_policy_form.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/policies/_components/add_policy_form.tsx @@ -1,10 +1,10 @@ import React, { useState, useEffect } from "react"; import { Form, Select, Modal, Divider, Typography, Tag, Alert, Radio } from "antd"; import { Button, TextInput, Textarea } from "@tremor/react"; -import { Policy, PolicyCreateRequest, PolicyUpdateRequest } from "./types"; -import { Guardrail } from "../guardrails/types"; -import { getResolvedGuardrails, modelAvailableCall } from "../networking"; -import NotificationsManager from "../molecules/notifications_manager"; +import { Policy, PolicyCreateRequest, PolicyUpdateRequest } from "@/components/policies/types"; +import { Guardrail } from "@/components/guardrails/types"; +import { getResolvedGuardrails, modelAvailableCall } from "@/components/networking"; +import NotificationsManager from "@/components/molecules/notifications_manager"; import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; const { Text } = Typography; diff --git a/ui/litellm-dashboard/src/components/policies/ai_suggestion_modal.tsx b/ui/litellm-dashboard/src/app/(dashboard)/policies/_components/ai_suggestion_modal.tsx similarity index 99% rename from ui/litellm-dashboard/src/components/policies/ai_suggestion_modal.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/policies/_components/ai_suggestion_modal.tsx index 3727db4d9cf..34204624702 100644 --- a/ui/litellm-dashboard/src/components/policies/ai_suggestion_modal.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/policies/_components/ai_suggestion_modal.tsx @@ -8,7 +8,12 @@ import { DownOutlined, RightOutlined, } from "@ant-design/icons"; -import { suggestPolicyTemplates, modelHubCall, testPolicyTemplate, enrichPolicyTemplateStream } from "../networking"; +import { + suggestPolicyTemplates, + modelHubCall, + testPolicyTemplate, + enrichPolicyTemplateStream, +} from "@/components/networking"; const { TextArea } = Input; const { Text } = Typography; diff --git a/ui/litellm-dashboard/src/components/policies/attachment_table.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/policies/_components/attachment_table.test.tsx similarity index 93% rename from ui/litellm-dashboard/src/components/policies/attachment_table.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/policies/_components/attachment_table.test.tsx index 099aa97a433..c53881e5cce 100644 --- a/ui/litellm-dashboard/src/components/policies/attachment_table.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/policies/_components/attachment_table.test.tsx @@ -1,10 +1,10 @@ import React from "react"; import { screen } from "@testing-library/react"; import userEvent from "@testing-library/user-event"; -import { renderWithProviders } from "../../../tests/test-utils"; +import { renderWithProviders } from "@/../tests/test-utils"; import { beforeEach, describe, expect, it, vi } from "vitest"; import AttachmentTable from "./attachment_table"; -import { PolicyAttachment } from "./types"; +import { PolicyAttachment } from "@/components/policies/types"; vi.mock("./impact_popover", () => ({ default: () => - - - { - e.stopPropagation(); - copyToClipboard(fullId); - }} - className="cursor-pointer text-gray-500 hover:text-blue-500 text-xs" - /> - -
- ); - }, + cell: (info: any) => , }, { header: "Model", @@ -162,26 +125,12 @@ const PromptTable: React.FC = ({ { header: "Created At", accessorKey: "created_at", - cell: ({ row }) => { - const prompt = row.original; - return ( - - {formatDate(prompt.created_at)} - - ); - }, + cell: ({ row }) => , }, { header: "Updated At", accessorKey: "updated_at", - cell: ({ row }) => { - const prompt = row.original; - return ( - - {formatDate(prompt.updated_at)} - - ); - }, + cell: ({ row }) => , }, { header: "Environment", diff --git a/ui/litellm-dashboard/src/app/(dashboard)/prompts/components/prompt_utils.tsx b/ui/litellm-dashboard/src/app/(dashboard)/prompts/_components/prompt_utils.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/prompts/components/prompt_utils.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/prompts/_components/prompt_utils.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/prompts/components/tool_modal.tsx b/ui/litellm-dashboard/src/app/(dashboard)/prompts/_components/tool_modal.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/prompts/components/tool_modal.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/prompts/_components/tool_modal.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/prompts/components/variable_textarea.tsx b/ui/litellm-dashboard/src/app/(dashboard)/prompts/_components/variable_textarea.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/prompts/components/variable_textarea.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/prompts/_components/variable_textarea.tsx diff --git a/ui/litellm-dashboard/src/app/(dashboard)/prompts/page.tsx b/ui/litellm-dashboard/src/app/(dashboard)/prompts/page.tsx index db0c511d083..7e251844a2c 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/prompts/page.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/prompts/page.tsx @@ -1,6 +1,6 @@ "use client"; -import PromptsPanel from "./components"; +import PromptsPanel from "./_components"; import { DeprecationBanner } from "@/components/DeprecationBanner"; import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; diff --git a/ui/litellm-dashboard/src/components/general_settings.tsx b/ui/litellm-dashboard/src/app/(dashboard)/router-settings/_components/general_settings.tsx similarity index 91% rename from ui/litellm-dashboard/src/components/general_settings.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/router-settings/_components/general_settings.tsx index 5b8dec39505..3955e80f5e9 100644 --- a/ui/litellm-dashboard/src/components/general_settings.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/router-settings/_components/general_settings.tsx @@ -4,7 +4,6 @@ import { Table, TableHead, TableRow, - Badge, TableHeaderCell, TableCell, TableBody, @@ -14,13 +13,14 @@ import { Switch, } from "@tremor/react"; import { TabPanel, TabPanels, TabGroup, TabList, Tab } from "@tremor/react"; -import { getGeneralSettingsCall, updateConfigFieldSetting, deleteConfigFieldSetting } from "./networking"; +import { getGeneralSettingsCall, updateConfigFieldSetting, deleteConfigFieldSetting } from "@/components/networking"; import { InputNumber } from "antd"; -import { TrashIcon, CheckCircleIcon } from "@heroicons/react/outline"; +import { TrashIcon } from "@heroicons/react/outline"; +import { StatusBadge } from "@/components/shared/table_cells"; -import RouterSettings from "./router_settings"; -import Fallbacks from "./Settings/RouterSettings/Fallbacks/Fallbacks"; -import RoutingGroups from "./routing_groups"; +import RouterSettings from "@/components/router_settings"; +import Fallbacks from "@/components/Settings/RouterSettings/Fallbacks/Fallbacks"; +import RoutingGroups from "@/components/routing_groups"; interface GeneralSettingsPageProps { accessToken: string | null; userRole: string | null; @@ -173,13 +173,11 @@ const GeneralSettings: React.FC = ({ accessToken, user {value.stored_in_db == true ? ( - - In DB - + ) : value.stored_in_db == false ? ( - In Config + ) : ( - Not Set + )} diff --git a/ui/litellm-dashboard/src/app/(dashboard)/router-settings/page.tsx b/ui/litellm-dashboard/src/app/(dashboard)/router-settings/page.tsx index 46029b529ec..90f41ac58a4 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/router-settings/page.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/router-settings/page.tsx @@ -1,6 +1,6 @@ "use client"; -import GeneralSettings from "@/components/general_settings"; +import GeneralSettings from "./_components/general_settings"; import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; export default function RouterSettingsPage() { diff --git a/ui/litellm-dashboard/src/app/(dashboard)/search-tools/_components/SearchToolColumn.tsx b/ui/litellm-dashboard/src/app/(dashboard)/search-tools/_components/SearchToolColumn.tsx index 198b3ea095f..3d9fdb2866e 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/search-tools/_components/SearchToolColumn.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/search-tools/_components/SearchToolColumn.tsx @@ -1,6 +1,7 @@ import { Tag } from "antd"; import { ColumnsType } from "antd/es/table"; import TableIconActionButton from "@/components/common_components/IconActionButton/TableIconActionButtons/TableIconActionButton"; +import { DateCell, IdCell } from "@/components/shared/table_cells"; import { SearchTool } from "./types"; export const searchToolColumns = ( @@ -20,14 +21,7 @@ export const searchToolColumns = ( return -; } - return ( - - ); + return ; }, }, { @@ -52,7 +46,7 @@ export const searchToolColumns = ( dataIndex: "created_at", key: "created_at", render: (_, tool) => { - return {tool.created_at ? new Date(tool.created_at).toLocaleDateString() : "-"}; + return ; }, }, { @@ -60,7 +54,7 @@ export const searchToolColumns = ( dataIndex: "updated_at", key: "updated_at", render: (_, tool) => { - return {tool.updated_at ? new Date(tool.updated_at).toLocaleDateString() : "-"}; + return ; }, }, { diff --git a/ui/litellm-dashboard/src/components/tag_management/TagTable.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/TagTable.test.tsx similarity index 79% rename from ui/litellm-dashboard/src/components/tag_management/TagTable.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/TagTable.test.tsx index a56721787d5..e9a35d961f9 100644 --- a/ui/litellm-dashboard/src/components/tag_management/TagTable.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/TagTable.test.tsx @@ -1,7 +1,8 @@ -import { render, screen } from "@testing-library/react"; +import { fireEvent, render, screen } from "@testing-library/react"; import { beforeEach, describe, expect, it, vi } from "vitest"; +import { formatCellDate } from "@/components/shared/table_cells"; import TagTable from "./TagTable"; -import { Tag } from "./types"; +import { Tag } from "@/components/tag_management/types"; describe("TagTable", () => { const mockOnEdit = vi.fn(); @@ -75,14 +76,21 @@ describe("TagTable", () => { it("should display formatted created date", () => { render(); - const formattedDate = new Date(mockTag.created_at).toLocaleDateString(); + const formattedDate = formatCellDate(new Date(mockTag.created_at), "date"); expect(screen.getByText(formattedDate)).toBeInTheDocument(); }); - it("should disable tag name button for dynamic spend tags", () => { + it("should call onSelectTag when tag name is clicked", () => { + render(); + fireEvent.click(screen.getByRole("button", { name: "test-tag" })); + expect(mockOnSelectTag).toHaveBeenCalledWith("test-tag"); + }); + + it("should render tag name as non-clickable for dynamic spend tags", () => { render(); - const tagButton = screen.getByRole("button", { name: "dynamic-spend-tag" }); - expect(tagButton).toBeDisabled(); + expect(screen.queryByRole("button", { name: "dynamic-spend-tag" })).not.toBeInTheDocument(); + fireEvent.click(screen.getByText("dynamic-spend-tag")); + expect(mockOnSelectTag).not.toHaveBeenCalled(); }); it("should disable edit icon for dynamic spend tags", () => { diff --git a/ui/litellm-dashboard/src/components/tag_management/TagTable.tsx b/ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/TagTable.tsx similarity index 90% rename from ui/litellm-dashboard/src/components/tag_management/TagTable.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/TagTable.tsx index ce28ac6e6f2..21bf4657a47 100644 --- a/ui/litellm-dashboard/src/components/tag_management/TagTable.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/TagTable.tsx @@ -7,21 +7,11 @@ import { SortingState, useReactTable, } from "@tanstack/react-table"; -import { - Badge, - Button, - Icon, - Table, - TableBody, - TableCell, - TableHead, - TableHeaderCell, - TableRow, - Text, -} from "@tremor/react"; +import { Badge, Icon, Table, TableBody, TableCell, TableHead, TableHeaderCell, TableRow, Text } from "@tremor/react"; import { Tooltip } from "antd"; import React from "react"; -import { Tag } from "./types"; +import { DateCell, IdCell } from "@/components/shared/table_cells"; +import { Tag } from "@/components/tag_management/types"; interface TagTableProps { data: Tag[]; @@ -45,21 +35,15 @@ const TagTable: React.FC = ({ data, onEdit, onDelete, onSelectTag const isDynamicSpendTag = tag.description === DYNAMIC_SPEND_TAG_DESCRIPTION; return (
- - - + />
); }, @@ -104,10 +88,7 @@ const TagTable: React.FC = ({ data, onEdit, onDelete, onSelectTag header: "Created", accessorKey: "created_at", sortingFn: "datetime", - cell: ({ row }) => { - const tag = row.original; - return {new Date(tag.created_at).toLocaleDateString()}; - }, + cell: ({ row }) => , }, { id: "actions", diff --git a/ui/litellm-dashboard/src/components/tag_management/components/CreateTagModal.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/components/CreateTagModal.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/components/tag_management/components/CreateTagModal.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/components/CreateTagModal.test.tsx diff --git a/ui/litellm-dashboard/src/components/tag_management/components/CreateTagModal.tsx b/ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/components/CreateTagModal.tsx similarity index 96% rename from ui/litellm-dashboard/src/components/tag_management/components/CreateTagModal.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/components/CreateTagModal.tsx index 4b1f7d34853..6fd50cbe823 100644 --- a/ui/litellm-dashboard/src/components/tag_management/components/CreateTagModal.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/components/CreateTagModal.tsx @@ -2,8 +2,8 @@ import { InfoCircleOutlined } from "@ant-design/icons"; import { Accordion, AccordionBody, AccordionHeader, Button, TextInput, Title } from "@tremor/react"; import { Form, Input, Modal, Select as Select2, Tooltip } from "antd"; import React from "react"; -import BudgetDurationDropdown from "../../common_components/budget_duration_dropdown"; -import NumericalInput from "../../shared/numerical_input"; +import BudgetDurationDropdown from "@/components/common_components/budget_duration_dropdown"; +import NumericalInput from "@/components/shared/numerical_input"; interface ModelInfo { model_name: string; diff --git a/ui/litellm-dashboard/src/components/tag_management/index.tsx b/ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/index.tsx similarity index 96% rename from ui/litellm-dashboard/src/components/tag_management/index.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/index.tsx index 1026f5842f9..043d229e3c1 100644 --- a/ui/litellm-dashboard/src/components/tag_management/index.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/index.tsx @@ -2,11 +2,11 @@ import React, { useState, useEffect } from "react"; import { Icon, Button, Col, Text, Grid } from "@tremor/react"; import { RefreshIcon } from "@heroicons/react/outline"; import TagInfoView from "./tag_info"; -import { modelInfoCall } from "../networking"; -import { tagCreateCall, tagListCall, tagDeleteCall } from "../networking"; -import { Tag } from "./types"; +import { modelInfoCall } from "@/components/networking"; +import { tagCreateCall, tagListCall, tagDeleteCall } from "@/components/networking"; +import { Tag } from "@/components/tag_management/types"; import TagTable from "./TagTable"; -import NotificationsManager from "../molecules/notifications_manager"; +import NotificationsManager from "@/components/molecules/notifications_manager"; import CreateTagModal from "./components/CreateTagModal"; interface ModelInfo { diff --git a/ui/litellm-dashboard/src/components/tag_management/tag_info.tsx b/ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/tag_info.tsx similarity index 95% rename from ui/litellm-dashboard/src/components/tag_management/tag_info.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/tag_info.tsx index 63725d5884f..19e08429730 100644 --- a/ui/litellm-dashboard/src/components/tag_management/tag_info.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/tag-management/_components/tag_info.tsx @@ -12,13 +12,13 @@ import { } from "@tremor/react"; import { Form, Input, Select as Select2, Tooltip } from "antd"; import { InfoCircleOutlined } from "@ant-design/icons"; -import { fetchUserModels } from "../organisms/create_key_button"; -import { getModelDisplayName } from "../key_team_helpers/fetch_available_models_team_key"; -import { tagInfoCall, tagUpdateCall } from "../networking"; -import { Tag } from "./types"; -import NotificationsManager from "../molecules/notifications_manager"; -import NumericalInput from "../shared/numerical_input"; -import BudgetDurationDropdown from "../common_components/budget_duration_dropdown"; +import { fetchUserModels } from "@/components/organisms/create_key_button"; +import { getModelDisplayName } from "@/components/key_team_helpers/fetch_available_models_team_key"; +import { tagInfoCall, tagUpdateCall } from "@/components/networking"; +import { Tag } from "@/components/tag_management/types"; +import NotificationsManager from "@/components/molecules/notifications_manager"; +import NumericalInput from "@/components/shared/numerical_input"; +import BudgetDurationDropdown from "@/components/common_components/budget_duration_dropdown"; import { copyToClipboard as utilCopyToClipboard } from "@/utils/dataUtils"; import { CheckIcon, CopyIcon } from "lucide-react"; import { Button as AntdButton } from "antd"; diff --git a/ui/litellm-dashboard/src/app/(dashboard)/tag-management/page.tsx b/ui/litellm-dashboard/src/app/(dashboard)/tag-management/page.tsx index 8a6ffd67388..5a1df715e76 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/tag-management/page.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/tag-management/page.tsx @@ -1,6 +1,6 @@ "use client"; -import TagManagement from "@/components/tag_management"; +import TagManagement from "./_components"; import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; export default function TagManagementPage() { diff --git a/ui/litellm-dashboard/src/app/(dashboard)/users/_components/BulkEditUsers.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/users/_components/BulkEditUsers.test.tsx index f16f5325952..e49ac854e6d 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/users/_components/BulkEditUsers.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/users/_components/BulkEditUsers.test.tsx @@ -91,7 +91,7 @@ describe("BulkEditUserModal", () => { it("should display budget information in table", () => { renderWithProviders(); - expect(screen.getByText("$50")).toBeInTheDocument(); + expect(screen.getByText("$50.00")).toBeInTheDocument(); expect(screen.getByText("Unlimited")).toBeInTheDocument(); }); diff --git a/ui/litellm-dashboard/src/app/(dashboard)/users/_components/BulkEditUsers.tsx b/ui/litellm-dashboard/src/app/(dashboard)/users/_components/BulkEditUsers.tsx index 9bef2f3a937..7ea53352807 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/users/_components/BulkEditUsers.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/users/_components/BulkEditUsers.tsx @@ -4,6 +4,7 @@ import { userBulkUpdateUserCall, teamBulkMemberAddCall, Member } from "@/compone import { UserEditView } from "./user_edit_view"; import NotificationsManager from "@/components/molecules/notifications_manager"; import MessageManager from "@/components/molecules/message_manager"; +import { MoneyCell } from "@/components/shared/table_cells"; const { Text, Title } = Typography; @@ -270,7 +271,7 @@ const BulkEditUserModal: React.FC = ({ key: "max_budget", width: "20%", render: (budget: number | null) => ( - {budget !== null ? `$${budget}` : "Unlimited"} + ), }, ]} diff --git a/ui/litellm-dashboard/src/app/(dashboard)/users/_components/view_users.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/users/_components/view_users.test.tsx index 241cda3464e..996fd58efc1 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/users/_components/view_users.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/users/_components/view_users.test.tsx @@ -158,7 +158,8 @@ describe("ViewUserDashboard", () => { expect( screen.getByText("Are you sure you want to delete this user? This action cannot be undone."), ).toBeInTheDocument(); - expect(screen.getByText("user-1")).toBeInTheDocument(); + const userIdInstances = screen.getAllByText("user-1"); + expect(userIdInstances.length).toBeGreaterThan(0); const emailInstances = screen.getAllByText("test@example.com"); expect(emailInstances.length).toBeGreaterThan(0); }); diff --git a/ui/litellm-dashboard/src/app/(dashboard)/users/_components/view_users/columns.tsx b/ui/litellm-dashboard/src/app/(dashboard)/users/_components/view_users/columns.tsx index deeaeeb25f4..fc680cb5b1b 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/users/_components/view_users/columns.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/users/_components/view_users/columns.tsx @@ -3,8 +3,7 @@ import { Badge, Grid, Icon } from "@tremor/react"; import { Tooltip, Checkbox, Tag } from "antd"; import { UserInfo } from "@/components/networking"; import { PencilAltIcon, TrashIcon, InformationCircleIcon, RefreshIcon } from "@heroicons/react/outline"; -import { CopyOutlined } from "@ant-design/icons"; -import { formatNumberWithCommas, copyToClipboard } from "@/utils/dataUtils"; +import { DateCell, IdCell, MoneyCell } from "@/components/shared/table_cells"; interface SelectionOptions { selectedUsers: UserInfo[]; @@ -29,24 +28,7 @@ export const columns = ( header: "User ID", accessorKey: "user_id", enableSorting: true, - cell: ({ row }) => ( -
- - {row.original.user_id ? `${row.original.user_id.slice(0, 7)}...` : "-"} - - {row.original.user_id && ( - - { - e.stopPropagation(); - copyToClipboard(row.original.user_id, "User ID copied to clipboard"); - }} - className="cursor-pointer text-gray-500 hover:text-blue-500 text-xs" - /> - - )} -
- ), + cell: ({ row }) => , }, { header: "Email", @@ -93,17 +75,13 @@ export const columns = ( header: "Spend (USD)", accessorKey: "spend", enableSorting: true, - cell: ({ row }) => ( - {row.original.spend ? formatNumberWithCommas(row.original.spend, 4) : "-"} - ), + cell: ({ row }) => , }, { header: "Budget (USD)", accessorKey: "max_budget", enableSorting: false, - cell: ({ row }) => ( - {row.original.max_budget !== null ? row.original.max_budget : "Unlimited"} - ), + cell: ({ row }) => , }, { header: () => ( @@ -142,21 +120,13 @@ export const columns = ( header: "Created At", accessorKey: "created_at", enableSorting: true, - cell: ({ row }) => ( - - {row.original.created_at ? new Date(row.original.created_at).toLocaleDateString() : "-"} - - ), + cell: ({ row }) => , }, { header: "Updated At", accessorKey: "updated_at", enableSorting: false, - cell: ({ row }) => ( - - {row.original.updated_at ? new Date(row.original.updated_at).toLocaleDateString() : "-"} - - ), + cell: ({ row }) => , }, { id: "actions", diff --git a/ui/litellm-dashboard/src/components/vector_store_management/CreateVectorStore.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/CreateVectorStore.test.tsx similarity index 97% rename from ui/litellm-dashboard/src/components/vector_store_management/CreateVectorStore.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/CreateVectorStore.test.tsx index 4069a7ec8ec..6c7a647e2d3 100644 --- a/ui/litellm-dashboard/src/components/vector_store_management/CreateVectorStore.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/CreateVectorStore.test.tsx @@ -1,15 +1,15 @@ import { render, screen, fireEvent, waitFor, act } from "@testing-library/react"; import { describe, it, expect, vi, beforeEach } from "vitest"; import CreateVectorStore from "./CreateVectorStore"; -import * as networking from "../networking"; +import * as networking from "@/components/networking"; // Mock the networking module -vi.mock("../networking", () => ({ +vi.mock("@/components/networking", () => ({ ragIngestCall: vi.fn(), })); // Mock NotificationsManager -vi.mock("../molecules/notifications_manager", () => ({ +vi.mock("@/components/molecules/notifications_manager", () => ({ default: { success: vi.fn(), fromBackend: vi.fn(), @@ -17,7 +17,7 @@ vi.mock("../molecules/notifications_manager", () => ({ })); // Mock vector_store_providers -vi.mock("../vector_store_providers", () => ({ +vi.mock("@/components/vector_store_providers", () => ({ VectorStoreProviders: { BEDROCK: "Amazon Bedrock", OPENAI: "OpenAI", diff --git a/ui/litellm-dashboard/src/components/vector_store_management/CreateVectorStore.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/CreateVectorStore.tsx similarity index 98% rename from ui/litellm-dashboard/src/components/vector_store_management/CreateVectorStore.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/CreateVectorStore.tsx index da88cda4e4d..b5f4b56986d 100644 --- a/ui/litellm-dashboard/src/components/vector_store_management/CreateVectorStore.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/CreateVectorStore.tsx @@ -4,8 +4,8 @@ import { Upload, Button, Select, Form, Alert, Tooltip, Input } from "antd"; import MessageManager from "@/components/molecules/message_manager"; import { InboxOutlined, InfoCircleOutlined } from "@ant-design/icons"; import type { UploadProps } from "antd"; -import { ragIngestCall } from "../networking"; -import { DocumentUpload, RAGIngestResponse } from "./types"; +import { ragIngestCall } from "@/components/networking"; +import { DocumentUpload, RAGIngestResponse } from "@/components/vector_store_management/types"; import DocumentsTable from "./DocumentsTable"; import { VectorStoreProviders, @@ -13,9 +13,9 @@ import { vectorStoreProviderMap, getProviderSpecificFields, VectorStoreFieldConfig, -} from "../vector_store_providers"; +} from "@/components/vector_store_providers"; import { resolveLogoSrc } from "@/lib/assetPaths"; -import NotificationsManager from "../molecules/notifications_manager"; +import NotificationsManager from "@/components/molecules/notifications_manager"; import S3VectorsConfig from "./S3VectorsConfig"; const { Dragger } = Upload; diff --git a/ui/litellm-dashboard/src/components/vector_store_management/DocumentsTable.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/DocumentsTable.test.tsx similarity index 97% rename from ui/litellm-dashboard/src/components/vector_store_management/DocumentsTable.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/DocumentsTable.test.tsx index 507ad0537d6..c2425c8d651 100644 --- a/ui/litellm-dashboard/src/components/vector_store_management/DocumentsTable.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/DocumentsTable.test.tsx @@ -1,7 +1,7 @@ import { render, screen, fireEvent, act } from "@testing-library/react"; import { describe, it, expect, vi } from "vitest"; import DocumentsTable from "./DocumentsTable"; -import { DocumentUpload } from "./types"; +import { DocumentUpload } from "@/components/vector_store_management/types"; // Mock antd message vi.mock("antd", async () => { diff --git a/ui/litellm-dashboard/src/components/vector_store_management/DocumentsTable.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/DocumentsTable.tsx similarity index 78% rename from ui/litellm-dashboard/src/components/vector_store_management/DocumentsTable.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/DocumentsTable.tsx index c8288172d5b..d4c45a6b751 100644 --- a/ui/litellm-dashboard/src/components/vector_store_management/DocumentsTable.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/DocumentsTable.tsx @@ -1,8 +1,9 @@ import React from "react"; -import { Table, Badge, Tooltip } from "antd"; +import { Table, Tooltip } from "antd"; import MessageManager from "@/components/molecules/message_manager"; import { EyeOutlined, CopyOutlined, DeleteOutlined } from "@ant-design/icons"; -import { DocumentUpload } from "./types"; +import { StatusBadge, type StatusTone } from "@/components/shared/table_cells"; +import { DocumentUpload } from "@/components/vector_store_management/types"; interface DocumentsTableProps { documents: DocumentUpload[]; @@ -16,15 +17,15 @@ const DocumentsTable: React.FC = ({ documents, onRemove }) }; const getStatusBadge = (status: DocumentUpload["status"]) => { - const statusConfig = { - uploading: { color: "blue", text: "Uploading" }, - done: { color: "green", text: "Ready" }, - error: { color: "red", text: "Error" }, - removed: { color: "default", text: "Removed" }, + const statusConfig: Record = { + uploading: { tone: "info", label: "Uploading" }, + done: { tone: "success", label: "Ready" }, + error: { tone: "error", label: "Error" }, + removed: { tone: "neutral", label: "Removed" }, }; - const config = statusConfig[status]; - return ; + const config: { tone: StatusTone; label: string } = statusConfig[status] ?? { tone: "neutral", label: status }; + return ; }; const formatFileSize = (bytes?: number) => { diff --git a/ui/litellm-dashboard/src/components/vector_store_management/S3VectorsConfig.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/S3VectorsConfig.test.tsx similarity index 100% rename from ui/litellm-dashboard/src/components/vector_store_management/S3VectorsConfig.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/S3VectorsConfig.test.tsx diff --git a/ui/litellm-dashboard/src/components/vector_store_management/S3VectorsConfig.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/S3VectorsConfig.tsx similarity index 100% rename from ui/litellm-dashboard/src/components/vector_store_management/S3VectorsConfig.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/S3VectorsConfig.tsx diff --git a/ui/litellm-dashboard/src/components/vector_store_management/TestVectorStoreTab.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/TestVectorStoreTab.test.tsx similarity index 97% rename from ui/litellm-dashboard/src/components/vector_store_management/TestVectorStoreTab.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/TestVectorStoreTab.test.tsx index ad6ef02e87b..c1322dced60 100644 --- a/ui/litellm-dashboard/src/components/vector_store_management/TestVectorStoreTab.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/TestVectorStoreTab.test.tsx @@ -1,7 +1,7 @@ import { render, screen, fireEvent } from "@testing-library/react"; import { describe, it, expect, vi } from "vitest"; import TestVectorStoreTab from "./TestVectorStoreTab"; -import { VectorStore } from "./types"; +import { VectorStore } from "@/components/vector_store_management/types"; // Mock VectorStoreTester component vi.mock("./VectorStoreTester", () => ({ diff --git a/ui/litellm-dashboard/src/components/vector_store_management/TestVectorStoreTab.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/TestVectorStoreTab.tsx similarity index 96% rename from ui/litellm-dashboard/src/components/vector_store_management/TestVectorStoreTab.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/TestVectorStoreTab.tsx index e96f149196e..3156da0b413 100644 --- a/ui/litellm-dashboard/src/components/vector_store_management/TestVectorStoreTab.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/TestVectorStoreTab.tsx @@ -1,7 +1,7 @@ import React, { useState } from "react"; import { Card, Select, Typography } from "antd"; import { VectorStoreTester } from "./VectorStoreTester"; -import { VectorStore } from "./types"; +import { VectorStore } from "@/components/vector_store_management/types"; const { Text, Title } = Typography; diff --git a/ui/litellm-dashboard/src/components/vector_store_management/VectorStoreForm.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/VectorStoreForm.test.tsx similarity index 88% rename from ui/litellm-dashboard/src/components/vector_store_management/VectorStoreForm.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/VectorStoreForm.test.tsx index 97d18e33640..e371cbfba42 100644 --- a/ui/litellm-dashboard/src/components/vector_store_management/VectorStoreForm.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/VectorStoreForm.test.tsx @@ -1,9 +1,9 @@ import { render, screen } from "@testing-library/react"; import { describe, expect, it, vi } from "vitest"; -import { CredentialItem } from "../networking"; +import { CredentialItem } from "@/components/networking"; import VectorStoreForm from "./VectorStoreForm"; -vi.mock("../networking"); +vi.mock("@/components/networking"); describe("VectorStoreForm", () => { it("should render the form when visible", () => { diff --git a/ui/litellm-dashboard/src/components/vector_store_management/VectorStoreForm.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/VectorStoreForm.tsx similarity index 98% rename from ui/litellm-dashboard/src/components/vector_store_management/VectorStoreForm.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/VectorStoreForm.tsx index 224b7e083a5..82417286738 100644 --- a/ui/litellm-dashboard/src/components/vector_store_management/VectorStoreForm.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/VectorStoreForm.tsx @@ -2,17 +2,17 @@ import React, { useState, useEffect } from "react"; import { TextInput, Button as TremorButton } from "@tremor/react"; import { Modal, Form, Select, Tooltip, Input, Alert } from "antd"; import { InfoCircleOutlined } from "@ant-design/icons"; -import { CredentialItem, vectorStoreCreateCall } from "../networking"; +import { CredentialItem, vectorStoreCreateCall } from "@/components/networking"; import { VectorStoreProviders, vectorStoreProviderLogoMap, vectorStoreProviderMap, getProviderSpecificFields, VectorStoreFieldConfig, -} from "../vector_store_providers"; +} from "@/components/vector_store_providers"; import { resolveLogoSrc } from "@/lib/assetPaths"; import { fetchAvailableModels, ModelGroup } from "@/components/llm_calls/fetch_models"; -import NotificationsManager from "../molecules/notifications_manager"; +import NotificationsManager from "@/components/molecules/notifications_manager"; interface VectorStoreFormProps { isVisible: boolean; diff --git a/ui/litellm-dashboard/src/components/vector_store_management/VectorStoreTable.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/VectorStoreTable.test.tsx similarity index 96% rename from ui/litellm-dashboard/src/components/vector_store_management/VectorStoreTable.test.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/VectorStoreTable.test.tsx index ac790122507..45ca40152fa 100644 --- a/ui/litellm-dashboard/src/components/vector_store_management/VectorStoreTable.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/VectorStoreTable.test.tsx @@ -2,17 +2,17 @@ import { render, screen } from "@testing-library/react"; import userEvent from "@testing-library/user-event"; import { beforeEach, describe, expect, it, vi } from "vitest"; import VectorStoreTable from "./VectorStoreTable"; -import { VectorStore } from "./types"; +import { VectorStore } from "@/components/vector_store_management/types"; // Mock dependencies const mockGetProviderLogoAndName = vi.fn(); const mockTableIconActionButton = vi.fn(); -vi.mock("../provider_info_helpers", () => ({ +vi.mock("@/components/provider_info_helpers", () => ({ getProviderLogoAndName: (...args: any[]) => mockGetProviderLogoAndName(...args), })); -vi.mock("../common_components/IconActionButton/TableIconActionButtons/TableIconActionButton", () => ({ +vi.mock("@/components/common_components/IconActionButton/TableIconActionButtons/TableIconActionButton", () => ({ default: (props: any) => { mockTableIconActionButton(props); return ( @@ -154,9 +154,8 @@ describe("VectorStoreTable", () => { it("should truncate long vector store IDs", () => { renderComponent(); - // Check that the truncated text is rendered (first 15 chars + ...) - const truncatedText = "very-long-vecto..."; - expect(screen.getByText(truncatedText)).toBeInTheDocument(); + const idButton = screen.getByText("very-long-vector-store-id-that-should-be-truncated"); + expect(idButton).toHaveClass("truncate", "max-w-[15ch]"); }); it("should make vector store ID clickable", async () => { @@ -245,13 +244,13 @@ describe("VectorStoreTable", () => { describe("Date Columns", () => { it("should render created at dates", () => { renderComponent(); - const dateElements = screen.getAllByText(/1\/\d+\/2024/); + const dateElements = screen.getAllByText(/Jan \d+, 2024/); expect(dateElements.length).toBe(6); // 3 created_at + 3 updated_at dates }); it("should render updated at dates", () => { renderComponent(); - const dateElements = screen.getAllByText(/1\/\d+\/2024/); + const dateElements = screen.getAllByText(/Jan \d+, 2024/); expect(dateElements.length).toBe(6); // 3 created_at + 3 updated_at dates }); }); diff --git a/ui/litellm-dashboard/src/components/vector_store_management/VectorStoreTable.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/VectorStoreTable.tsx similarity index 85% rename from ui/litellm-dashboard/src/components/vector_store_management/VectorStoreTable.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/VectorStoreTable.tsx index 180c2485ab5..074af873601 100644 --- a/ui/litellm-dashboard/src/components/vector_store_management/VectorStoreTable.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/VectorStoreTable.tsx @@ -10,9 +10,10 @@ import { import { Table, TableBody, TableCell, TableHead, TableHeaderCell, TableRow } from "@tremor/react"; import { Tooltip } from "antd"; import React from "react"; -import TableIconActionButton from "../common_components/IconActionButton/TableIconActionButtons/TableIconActionButton"; -import { getProviderLogoAndName } from "../provider_info_helpers"; -import { VectorStore } from "./types"; +import { DateCell, IdCell } from "@/components/shared/table_cells"; +import TableIconActionButton from "@/components/common_components/IconActionButton/TableIconActionButtons/TableIconActionButton"; +import { getProviderLogoAndName } from "@/components/provider_info_helpers"; +import { VectorStore } from "@/components/vector_store_management/types"; interface VectorStoreTableProps { data: VectorStore[]; @@ -28,19 +29,7 @@ const VectorStoreTable: React.FC = ({ data, onView, onEdi { header: "Vector Store ID", accessorKey: "vector_store_id", - cell: ({ row }) => { - const vectorStore = row.original; - return ( - - ); - }, + cell: ({ row }) => , }, { header: "Name", @@ -109,19 +98,13 @@ const VectorStoreTable: React.FC = ({ data, onView, onEdi header: "Created At", accessorKey: "created_at", sortingFn: "datetime", - cell: ({ row }) => { - const vectorStore = row.original; - return {new Date(vectorStore.created_at).toLocaleDateString()}; - }, + cell: ({ row }) => , }, { header: "Updated At", accessorKey: "updated_at", sortingFn: "datetime", - cell: ({ row }) => { - const vectorStore = row.original; - return {new Date(vectorStore.updated_at).toLocaleDateString()}; - }, + cell: ({ row }) => , }, { id: "actions", diff --git a/ui/litellm-dashboard/src/components/vector_store_management/VectorStoreTester.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/VectorStoreTester.tsx similarity index 98% rename from ui/litellm-dashboard/src/components/vector_store_management/VectorStoreTester.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/VectorStoreTester.tsx index a90f49ea5b3..fa4587b526c 100644 --- a/ui/litellm-dashboard/src/components/vector_store_management/VectorStoreTester.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/VectorStoreTester.tsx @@ -2,8 +2,8 @@ import React, { useState } from "react"; import { Button, Input, Card, Typography, Spin, Divider } from "antd"; import MessageManager from "@/components/molecules/message_manager"; import { SendOutlined, DatabaseOutlined, LoadingOutlined, DownOutlined, RightOutlined } from "@ant-design/icons"; -import { vectorStoreSearchCall } from "../networking"; -import NotificationsManager from "../molecules/notifications_manager"; +import { vectorStoreSearchCall } from "@/components/networking"; +import NotificationsManager from "@/components/molecules/notifications_manager"; const { TextArea } = Input; const { Text, Title } = Typography; diff --git a/ui/litellm-dashboard/src/components/vector_store_management/index.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/index.tsx similarity index 94% rename from ui/litellm-dashboard/src/components/vector_store_management/index.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/index.tsx index 354a79e4a2a..fd02effc972 100644 --- a/ui/litellm-dashboard/src/components/vector_store_management/index.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/index.tsx @@ -12,16 +12,21 @@ import { TabPanel, } from "@tremor/react"; import { RefreshIcon } from "@heroicons/react/outline"; -import { vectorStoreListCall, vectorStoreDeleteCall, credentialListCall, CredentialItem } from "../networking"; -import { VectorStore } from "./types"; +import { + vectorStoreListCall, + vectorStoreDeleteCall, + credentialListCall, + CredentialItem, +} from "@/components/networking"; +import { VectorStore } from "@/components/vector_store_management/types"; import VectorStoreTable from "./VectorStoreTable"; import VectorStoreForm from "./VectorStoreForm"; -import DeleteResourceModal from "../common_components/DeleteResourceModal"; +import DeleteResourceModal from "@/components/common_components/DeleteResourceModal"; import VectorStoreInfoView from "./vector_store_info"; import CreateVectorStore from "./CreateVectorStore"; import TestVectorStoreTab from "./TestVectorStoreTab"; import { isAdminRole } from "@/utils/roles"; -import NotificationsManager from "../molecules/notifications_manager"; +import NotificationsManager from "@/components/molecules/notifications_manager"; interface VectorStoreProps { accessToken: string | null; diff --git a/ui/litellm-dashboard/src/components/vector_store_management/vector_store_info.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/vector_store_info.tsx similarity index 97% rename from ui/litellm-dashboard/src/components/vector_store_management/vector_store_info.tsx rename to ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/vector_store_info.tsx index 1929224d01d..7c27b347eb1 100644 --- a/ui/litellm-dashboard/src/components/vector_store_management/vector_store_info.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/_components/vector_store_info.tsx @@ -3,12 +3,17 @@ import { Card, Text, Title, Button, Badge, TabGroup, TabList, Tab, TabPanels, Ta import { Form, Input, Select as Select2, Tooltip, Button as AntButton } from "antd"; import { InfoCircleOutlined } from "@ant-design/icons"; import { ArrowLeftIcon } from "@heroicons/react/outline"; -import { vectorStoreInfoCall, vectorStoreUpdateCall, credentialListCall, CredentialItem } from "../networking"; -import { VectorStore } from "./types"; -import { Providers, providerLogoMap, provider_map } from "../provider_info_helpers"; +import { + vectorStoreInfoCall, + vectorStoreUpdateCall, + credentialListCall, + CredentialItem, +} from "@/components/networking"; +import { VectorStore } from "@/components/vector_store_management/types"; +import { Providers, providerLogoMap, provider_map } from "@/components/provider_info_helpers"; import { resolveLogoSrc } from "@/lib/assetPaths"; import VectorStoreTester from "./VectorStoreTester"; -import NotificationsManager from "../molecules/notifications_manager"; +import NotificationsManager from "@/components/molecules/notifications_manager"; interface VectorStoreInfoViewProps { vectorStoreId: string; diff --git a/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/page.tsx b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/page.tsx index ba280c1c313..34759152d11 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/page.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/vector-stores/page.tsx @@ -1,6 +1,6 @@ "use client"; -import VectorStoreManagement from "@/components/vector_store_management"; +import VectorStoreManagement from "./_components"; import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; export default function VectorStores() { diff --git a/ui/litellm-dashboard/src/components/AIHub/AgentHubTableColumns.tsx b/ui/litellm-dashboard/src/components/AIHub/AgentHubTableColumns.tsx index 09b1c147615..ae1a19ff95d 100644 --- a/ui/litellm-dashboard/src/components/AIHub/AgentHubTableColumns.tsx +++ b/ui/litellm-dashboard/src/components/AIHub/AgentHubTableColumns.tsx @@ -2,6 +2,7 @@ import { ColumnDef } from "@tanstack/react-table"; import { Button, Badge, Text } from "@tremor/react"; import { Tooltip, Tag } from "antd"; import { CopyOutlined, InfoCircleOutlined } from "@ant-design/icons"; +import { StatusBadge } from "@/components/shared/table_cells"; export interface AgentHubData { agent_id?: string; @@ -194,17 +195,9 @@ export const getAgentHubTableColumns = ( return publicA - publicB; }, cell: ({ row }) => { - const agent = row.original; + const isPublic = row.original.is_public === true; - return agent.is_public === true ? ( - - Yes - - ) : ( - - No - - ); + return ; }, meta: { className: "hidden md:table-cell", diff --git a/ui/litellm-dashboard/src/app/(dashboard)/api-reference/components/CodeBlock.tsx b/ui/litellm-dashboard/src/components/CodeBlock.tsx similarity index 100% rename from ui/litellm-dashboard/src/app/(dashboard)/api-reference/components/CodeBlock.tsx rename to ui/litellm-dashboard/src/components/CodeBlock.tsx diff --git a/ui/litellm-dashboard/src/components/DeletedKeysPage/DeletedKeysTable/DeletedKeysTable.tsx b/ui/litellm-dashboard/src/components/DeletedKeysPage/DeletedKeysTable/DeletedKeysTable.tsx index bc52bbbe062..d4a120d0589 100644 --- a/ui/litellm-dashboard/src/components/DeletedKeysPage/DeletedKeysTable/DeletedKeysTable.tsx +++ b/ui/litellm-dashboard/src/components/DeletedKeysPage/DeletedKeysTable/DeletedKeysTable.tsx @@ -1,5 +1,5 @@ "use client"; -import { formatNumberWithCommas } from "@/utils/dataUtils"; +import { DateCell, IdCell, MoneyCell } from "@/components/shared/table_cells"; import { ChevronDownIcon, ChevronUpIcon, SwitchVerticalIcon } from "@heroicons/react/outline"; import { ColumnDef, @@ -59,14 +59,7 @@ export function DeletedKeysTable({ header: "Key ID", size: 150, maxSize: 250, - cell: (info) => { - const value = info.getValue() as string; - return ( - - {value || "-"} - - ); - }, + cell: (info) => , }, { id: "key_alias", @@ -100,9 +93,7 @@ export function DeletedKeysTable({ header: "Spend (USD)", size: 100, maxSize: 140, - cell: (info) => ( - {formatNumberWithCommas(info.getValue() as number, 4)} - ), + cell: (info) => , }, { id: "max_budget", @@ -110,14 +101,9 @@ export function DeletedKeysTable({ header: "Budget (USD)", size: 110, maxSize: 150, - cell: (info) => { - const maxBudget = info.getValue() as number | null; - return ( - - {maxBudget === null ? "Unlimited" : `$${formatNumberWithCommas(maxBudget)}`} - - ); - }, + cell: (info) => ( + + ), }, { id: "user_email", @@ -140,14 +126,7 @@ export function DeletedKeysTable({ header: "User ID", size: 120, maxSize: 200, - cell: (info) => { - const userId = info.getValue() as string | null; - return ( - - {userId || "-"} - - ); - }, + cell: (info) => , }, { id: "created_at", @@ -155,12 +134,7 @@ export function DeletedKeysTable({ header: "Created At", size: 120, maxSize: 140, - cell: (info) => { - const value = info.getValue(); - return ( - {value ? new Date(value as string).toLocaleDateString() : "-"} - ); - }, + cell: (info) => , }, { id: "created_by", @@ -183,10 +157,9 @@ export function DeletedKeysTable({ header: "Deleted At", size: 120, maxSize: 140, - cell: (info) => { - const value = (info.row.original as any).deleted_at as string | null | undefined; - return {value ? new Date(value).toLocaleDateString() : "-"}; - }, + cell: (info) => ( + + ), }, { id: "deleted_by", diff --git a/ui/litellm-dashboard/src/components/DeletedTeamsPage/DeletedTeamsTable/DeletedTeamsTable.tsx b/ui/litellm-dashboard/src/components/DeletedTeamsPage/DeletedTeamsTable/DeletedTeamsTable.tsx index 57260f6bb2d..ddfd5cf73b6 100644 --- a/ui/litellm-dashboard/src/components/DeletedTeamsPage/DeletedTeamsTable/DeletedTeamsTable.tsx +++ b/ui/litellm-dashboard/src/components/DeletedTeamsPage/DeletedTeamsTable/DeletedTeamsTable.tsx @@ -1,5 +1,5 @@ "use client"; -import { formatNumberWithCommas } from "@/utils/dataUtils"; +import { DateCell, IdCell, MoneyCell } from "@/components/shared/table_cells"; import { ChevronDownIcon, ChevronUpIcon, SwitchVerticalIcon } from "@heroicons/react/outline"; import { ColumnDef, @@ -51,14 +51,7 @@ export function DeletedTeamsTable({ teams, isLoading, isFetching }: DeletedTeams header: "Team ID", size: 150, maxSize: 250, - cell: (info) => { - const value = info.getValue() as string; - return ( - - {value || "-"} - - ); - }, + cell: (info) => , }, { id: "created_at", @@ -66,12 +59,7 @@ export function DeletedTeamsTable({ teams, isLoading, isFetching }: DeletedTeams header: "Created", size: 120, maxSize: 140, - cell: (info) => { - const value = info.getValue(); - return ( - {value ? new Date(value as string).toLocaleDateString() : "-"} - ); - }, + cell: (info) => , }, { id: "spend", @@ -79,12 +67,7 @@ export function DeletedTeamsTable({ teams, isLoading, isFetching }: DeletedTeams header: "Spend (USD)", size: 100, maxSize: 140, - cell: (info) => { - const spend = (info.row.original as any).spend as number | undefined; - return ( - {spend !== undefined ? formatNumberWithCommas(spend, 4) : "-"} - ); - }, + cell: (info) => , }, { id: "max_budget", @@ -92,14 +75,9 @@ export function DeletedTeamsTable({ teams, isLoading, isFetching }: DeletedTeams header: "Budget (USD)", size: 110, maxSize: 150, - cell: (info) => { - const maxBudget = info.getValue() as number | null; - return ( - - {maxBudget === null || maxBudget === undefined ? "No limit" : `$${formatNumberWithCommas(maxBudget)}`} - - ); - }, + cell: (info) => ( + + ), }, { id: "models", @@ -148,14 +126,7 @@ export function DeletedTeamsTable({ teams, isLoading, isFetching }: DeletedTeams header: "Organization", size: 150, maxSize: 200, - cell: (info) => { - const value = info.getValue() as string; - return ( - - {value || "-"} - - ); - }, + cell: (info) => , }, { id: "deleted_at", @@ -163,10 +134,7 @@ export function DeletedTeamsTable({ teams, isLoading, isFetching }: DeletedTeams header: "Deleted At", size: 120, maxSize: 140, - cell: (info) => { - const value = (info.row.original as any).deleted_at as string | null | undefined; - return {value ? new Date(value).toLocaleDateString() : "-"}; - }, + cell: (info) => , }, { id: "deleted_by", diff --git a/ui/litellm-dashboard/src/components/LicenseExpiryBanner.test.tsx b/ui/litellm-dashboard/src/components/LicenseExpiryBanner.test.tsx new file mode 100644 index 00000000000..d6b419ace7c --- /dev/null +++ b/ui/litellm-dashboard/src/components/LicenseExpiryBanner.test.tsx @@ -0,0 +1,90 @@ +import React from "react"; +import { describe, it, expect, beforeEach, afterEach } from "vitest"; +import { render, screen, fireEvent } from "@testing-library/react"; +import { LicenseExpiryBannerView } from "./LicenseExpiryBanner"; +import { LicenseInfo } from "./networking"; + +const daysFromNow = (n: number): string => { + const date = new Date(); + date.setUTCDate(date.getUTCDate() + n); + return date.toISOString().slice(0, 10); +}; + +const licenseWith = (expiration_date: string | null): LicenseInfo => ({ + has_license: expiration_date !== null, + license_type: expiration_date !== null ? "enterprise" : "community", + expiration_date, + allowed_features: [], + limits: { max_users: null, max_teams: null }, +}); + +describe("LicenseExpiryBannerView", () => { + beforeEach(() => { + sessionStorage.clear(); + }); + + afterEach(() => { + sessionStorage.clear(); + }); + + it("renders nothing when there is no license info", () => { + const { container } = render(); + expect(container).toBeEmptyDOMElement(); + }); + + it("renders nothing when expiration_date is null (community or remote-validated)", () => { + const { container } = render(); + expect(container).toBeEmptyDOMElement(); + }); + + it("renders nothing when expiry is more than 30 days out", () => { + const { container } = render(); + expect(container).toBeEmptyDOMElement(); + }); + + it("shows a dismissible amber warning within 30 days", () => { + const { container } = render(); + expect(screen.getByText(/expires in 20 days/)).toBeInTheDocument(); + expect(container.querySelector(".ant-alert-warning")).toBeInTheDocument(); + expect(screen.queryByRole("button")).toBeInTheDocument(); + expect(screen.getByRole("link", { name: "sales@berri.ai" })).toHaveAttribute("href", "mailto:sales@berri.ai"); + }); + + it("shows a non-dismissible red critical alert within 7 days", () => { + const { container } = render(); + expect(screen.getByText(/expires in 5 days/)).toBeInTheDocument(); + expect(container.querySelector(".ant-alert-error")).toBeInTheDocument(); + expect(screen.queryByRole("button")).not.toBeInTheDocument(); + }); + + it("says 'expires today' on the expiration day", () => { + render(); + expect(screen.getByText(/expires today/)).toBeInTheDocument(); + }); + + it("shows a non-dismissible red expired alert stating features are disabled", () => { + const { container } = render(); + expect(screen.getByText(/expired on/)).toBeInTheDocument(); + expect(screen.getByText(/features are now disabled/i)).toBeInTheDocument(); + expect(container.querySelector(".ant-alert-error")).toBeInTheDocument(); + expect(screen.queryByRole("button")).not.toBeInTheDocument(); + }); + + it("hides the warning after dismissal and stays hidden within the session", () => { + const expiration = daysFromNow(20); + const { unmount } = render(); + fireEvent.click(screen.getByRole("button")); + expect(screen.queryByText(/expires in 20 days/)).not.toBeInTheDocument(); + + unmount(); + render(); + expect(screen.queryByText(/expires in 20 days/)).not.toBeInTheDocument(); + }); + + it("still shows a critical alert even when its date was previously dismissed", () => { + const expiration = daysFromNow(5); + sessionStorage.setItem(`litellm:licenseExpiryBannerDismissed:${expiration}`, "true"); + render(); + expect(screen.getByText(/expires in 5 days/)).toBeInTheDocument(); + }); +}); diff --git a/ui/litellm-dashboard/src/components/LicenseExpiryBanner.tsx b/ui/litellm-dashboard/src/components/LicenseExpiryBanner.tsx new file mode 100644 index 00000000000..c3b20b5fac0 --- /dev/null +++ b/ui/litellm-dashboard/src/components/LicenseExpiryBanner.tsx @@ -0,0 +1,95 @@ +"use client"; + +import React, { useState } from "react"; +import { Alert } from "antd"; +import { LicenseInfo } from "@/components/networking"; +import { useLicenseInfo } from "@/app/(dashboard)/hooks/license/useLicenseInfo"; +import { formatExpiryDate, getDaysUntilExpiration, getLicenseExpiryTier } from "@/utils/licenseUtils"; + +const DISMISS_KEY_PREFIX = "litellm:licenseExpiryBannerDismissed:"; +const SALES_EMAIL = "sales@berri.ai"; + +const salesLink = {SALES_EMAIL}; + +interface LicenseExpiryBannerProps { + accessToken: string | null; +} + +interface LicenseExpiryBannerViewProps { + licenseInfo: LicenseInfo | null; +} + +const describeCountdown = (days: number): string => { + if (days <= 0) { + return "expires today"; + } + if (days === 1) { + return "expires in 1 day"; + } + return `expires in ${days} days`; +}; + +const expiryDescription = (tier: "warning" | "critical" | "expired"): React.ReactNode => { + if (tier === "expired") { + return <>Enterprise features are now disabled. Reach out to {salesLink} to restore access; + } + if (tier === "critical") { + return <>Renew now to avoid losing enterprise features. Reach out to {salesLink}; + } + return <>Renew before it lapses to keep enterprise features. Reach out to {salesLink}; +}; + +export const LicenseExpiryBannerView: React.FC = ({ licenseInfo }) => { + const [locallyDismissed, setLocallyDismissed] = useState(false); + + const expirationDate = licenseInfo?.expiration_date ?? null; + const tier = getLicenseExpiryTier(expirationDate); + const days = getDaysUntilExpiration(expirationDate); + + if (expirationDate === null || tier === "none" || days === null) { + return null; + } + + const isDismissible = tier === "warning"; + const dismissKey = `${DISMISS_KEY_PREFIX}${expirationDate}`; + const previouslyDismissed = + isDismissible && typeof window !== "undefined" ? sessionStorage.getItem(dismissKey) === "true" : false; + + if (isDismissible && (locallyDismissed || previouslyDismissed)) { + return null; + } + + const formattedDate = formatExpiryDate(expirationDate); + + const message = + tier === "expired" + ? `Your LiteLLM Enterprise license expired on ${formattedDate}` + : `Your LiteLLM Enterprise license ${describeCountdown(days)} (${formattedDate})`; + + const description = expiryDescription(tier); + + const handleClose = () => { + if (typeof window !== "undefined") { + sessionStorage.setItem(dismissKey, "true"); + } + setLocallyDismissed(true); + }; + + return ( + + ); +}; + +export const LicenseExpiryBanner: React.FC = ({ accessToken }) => { + const { data } = useLicenseInfo(accessToken); + return ; +}; diff --git a/ui/litellm-dashboard/src/components/OldTeams.tsx b/ui/litellm-dashboard/src/components/OldTeams.tsx index c2cb3d4948e..e83d1acf4e5 100644 --- a/ui/litellm-dashboard/src/components/OldTeams.tsx +++ b/ui/litellm-dashboard/src/components/OldTeams.tsx @@ -31,6 +31,7 @@ import type { SorterResult } from "antd/es/table/interface"; import { KeyIcon, LayersIcon, SearchIcon, UsersIcon } from "lucide-react"; import React, { useEffect, useMemo, useRef, useState } from "react"; import { AntDLoadingSpinner } from "@/components/ui/AntDLoadingSpinner"; +import { DateCell, IdCell } from "@/components/shared/table_cells"; import OrganizationDropdown from "./common_components/OrganizationDropdown"; import TableIconActionButton from "./common_components/IconActionButton/TableIconActionButtons/TableIconActionButton"; import { teamListCall as v2TeamListCall, type TeamsResponse } from "@/app/(dashboard)/hooks/teams/useTeams"; @@ -670,18 +671,8 @@ const Teams: React.FC = ({ accessToken, userID, userRole, premiumUser key: "team_id", width: 170, ellipsis: true, - render: (id: string, record: Team) => ( - - setSelectedTeamId(record.team_id)} - data-testid="team-id-cell" - > - {id} - - + render: (id: string) => ( + setSelectedTeamId(teamId)} dataTestId="team-id-cell" /> ), }, { @@ -797,13 +788,7 @@ const Teams: React.FC = ({ accessToken, userID, userRole, premiumUser width: 130, ellipsis: true, sorter: true, - render: (date: string | undefined) => ( - - {date - ? new Date(date).toLocaleDateString(undefined, { year: "numeric", month: "short", day: "numeric" }) - : "—"} - - ), + render: (date: string | undefined) => , }, { title: "Actions", diff --git a/ui/litellm-dashboard/src/components/Settings/LoggingAndAlerts/LoggingCallbacks/LoggingCallbacksTable.tsx b/ui/litellm-dashboard/src/components/Settings/LoggingAndAlerts/LoggingCallbacks/LoggingCallbacksTable.tsx index 70ec6599ca2..4f1889cbc99 100644 --- a/ui/litellm-dashboard/src/components/Settings/LoggingAndAlerts/LoggingCallbacks/LoggingCallbacksTable.tsx +++ b/ui/litellm-dashboard/src/components/Settings/LoggingAndAlerts/LoggingCallbacks/LoggingCallbacksTable.tsx @@ -3,6 +3,7 @@ import type { TableProps } from "antd"; import { Table } from "antd"; import Title from "antd/es/typography/Title"; import React from "react"; +import { StatusBadge, type StatusTone } from "@/components/shared/table_cells"; import TableIconActionButton from "../../../common_components/IconActionButton/TableIconActionButtons/TableIconActionButton"; import { AlertingObject } from "./types"; @@ -61,17 +62,8 @@ export const LoggingCallbacksTable: React.FC = ({ // and server-fetched rows both render correctly. const mode = record.type || record.mode || "success"; const label = CALLBACK_MODES.find((m) => m.value === mode)?.label || mode; - const badgeClass = - mode === "success" - ? "bg-green-100 text-green-800" - : mode === "failure" - ? "bg-red-100 text-red-800" - : "bg-blue-100 text-blue-800"; - return ( - - {label} - - ); + const tone: StatusTone = mode === "success" ? "success" : mode === "failure" ? "error" : "info"; + return ; }, width: 240, }, diff --git a/ui/litellm-dashboard/src/components/ToolPolicies.tsx b/ui/litellm-dashboard/src/components/ToolPolicies.tsx index 4bd028f0c8f..4468334f813 100644 --- a/ui/litellm-dashboard/src/components/ToolPolicies.tsx +++ b/ui/litellm-dashboard/src/components/ToolPolicies.tsx @@ -3,7 +3,7 @@ import React, { useCallback, useDeferredValue, useEffect, useMemo, useState } from "react"; import { Button, Switch, Tooltip } from "antd"; import { Table, TableHead, TableHeaderCell, TableBody, TableRow, TableCell } from "@tremor/react"; -import { TimeCell } from "./view_logs/time_cell"; +import { DateCell, IdCell } from "@/components/shared/table_cells"; import type { SortState } from "./common_components/TableHeaderSortDropdown/TableHeaderSortDropdown"; import { TableHeaderSortDropdown } from "./common_components/TableHeaderSortDropdown/TableHeaderSortDropdown"; import FilterComponent, { FilterOption } from "./molecules/filter"; @@ -461,7 +461,7 @@ export const ToolPolicies: React.FC = ({ accessToken, onSelec paginated.map((tool) => ( - +
- - {tool.team_id ?? "-"} - + - - - {tool.key_hash ?? "-"} - - + diff --git a/ui/litellm-dashboard/src/components/UsageIndicator.test.tsx b/ui/litellm-dashboard/src/components/UsageIndicator.test.tsx index 8c7c15bc5a5..ad27fbcd74f 100644 --- a/ui/litellm-dashboard/src/components/UsageIndicator.test.tsx +++ b/ui/litellm-dashboard/src/components/UsageIndicator.test.tsx @@ -2,6 +2,7 @@ import React from "react"; import { describe, it, expect, vi, beforeEach } from "vitest"; import { render, screen, waitFor } from "@testing-library/react"; import userEvent from "@testing-library/user-event"; +import { QueryClient, QueryClientProvider } from "@tanstack/react-query"; import UsageIndicator from "./UsageIndicator"; vi.mock("./networking", () => ({ @@ -17,6 +18,11 @@ import { getRemainingUsers } from "./networking"; const mockGetRemainingUsers = vi.mocked(getRemainingUsers); +const renderWithClient = (ui: React.ReactElement) => { + const queryClient = new QueryClient({ defaultOptions: { queries: { retry: false } } }); + return render({ui}); +}; + const DEFAULT_USAGE_DATA = { total_users: 100, total_users_used: 1, @@ -33,7 +39,7 @@ describe("UsageIndicator", () => { }); it("should render when given access token and usage data loads", async () => { - render(); + renderWithClient(); await screen.findByText("Usage"); @@ -41,7 +47,7 @@ describe("UsageIndicator", () => { }); it("should not show Near limit when users usage is below 80% (1/100 -> 1%)", async () => { - render(); + renderWithClient(); await screen.findByText("Usage"); @@ -58,7 +64,7 @@ describe("UsageIndicator", () => { total_users_remaining: null, }); - render(); + renderWithClient(); await waitFor(() => { expect(screen.queryByText("Usage")).not.toBeInTheDocument(); @@ -76,7 +82,7 @@ describe("UsageIndicator", () => { total_teams_remaining: 1, }); - render(); + renderWithClient(); await screen.findByText("Usage"); @@ -94,7 +100,7 @@ describe("UsageIndicator", () => { total_teams_remaining: null, }); - render(); + renderWithClient(); await screen.findByText("Usage"); @@ -112,7 +118,7 @@ describe("UsageIndicator", () => { total_teams_remaining: -2, }); - render(); + renderWithClient(); await screen.findByText("Usage"); @@ -121,7 +127,7 @@ describe("UsageIndicator", () => { }); it("should render nothing when accessToken is null", () => { - render(); + renderWithClient(); expect(mockGetRemainingUsers).not.toHaveBeenCalled(); expect(screen.queryByText("Usage")).not.toBeInTheDocument(); @@ -131,7 +137,7 @@ describe("UsageIndicator", () => { const { useDisableUsageIndicator } = await import("@/app/(dashboard)/hooks/useDisableUsageIndicator"); (useDisableUsageIndicator as ReturnType).mockReturnValue(true); - render(); + renderWithClient(); await waitFor(() => { expect(screen.queryByText("Usage")).not.toBeInTheDocument(); @@ -143,7 +149,7 @@ describe("UsageIndicator", () => { it("should show Loading while fetching", () => { mockGetRemainingUsers.mockImplementation(() => new Promise(() => {})); - render(); + renderWithClient(); expect(screen.getByText("Loading...")).toBeInTheDocument(); }); @@ -152,7 +158,7 @@ describe("UsageIndicator", () => { const consoleSpy = vi.spyOn(console, "error").mockImplementation(() => {}); mockGetRemainingUsers.mockRejectedValue(new Error("Network error")); - render(); + renderWithClient(); expect(await screen.findByText("Failed to load usage data")).toBeInTheDocument(); @@ -161,7 +167,7 @@ describe("UsageIndicator", () => { it("should minimize when user clicks minimize button", async () => { const user = userEvent.setup(); - render(); + renderWithClient(); await screen.findByText("Usage"); @@ -174,7 +180,7 @@ describe("UsageIndicator", () => { it("should restore from minimized when user clicks restore button", async () => { const user = userEvent.setup(); - render(); + renderWithClient(); await screen.findByText("Usage"); diff --git a/ui/litellm-dashboard/src/components/UsageIndicator.tsx b/ui/litellm-dashboard/src/components/UsageIndicator.tsx index 030df228de0..6e7b5e9ec60 100644 --- a/ui/litellm-dashboard/src/components/UsageIndicator.tsx +++ b/ui/litellm-dashboard/src/components/UsageIndicator.tsx @@ -12,9 +12,11 @@ import { Users, } from "lucide-react"; import { useEffect, useState } from "react"; -import { getRemainingUsers, getLicenseInfo, LicenseInfo } from "./networking"; +import { getRemainingUsers } from "./networking"; import { cn } from "@/lib/cva.config"; +import { getDaysUntilExpiration } from "@/utils/licenseUtils"; +import { useLicenseInfo } from "@/app/(dashboard)/hooks/license/useLicenseInfo"; interface UsageIndicatorProps { accessToken: string | null; @@ -30,17 +32,6 @@ interface UsageData { total_teams_remaining: number | null; } -// Calculate days until expiration -const getDaysUntilExpiration = (expirationDate: string | null): number | null => { - if (!expirationDate) return null; - const expDate = new Date(expirationDate + "T00:00:00Z"); // Force UTC midnight - const now = new Date(); - now.setHours(0, 0, 0, 0); // Normalize to local midnight - const diffTime = expDate.getTime() - now.getTime(); - const diffDays = Math.ceil(diffTime / (1000 * 60 * 60 * 24)); - return diffDays; -}; - // Format expiration for display const formatExpirationDisplay = (daysRemaining: number | null): string => { if (daysRemaining === null) return "No expiration"; @@ -58,10 +49,11 @@ export default function UsageIndicator({ accessToken, width = 220 }: UsageIndica const [isExpanded, setIsExpanded] = useState(false); const [isMinimized, setIsMinimized] = useState(false); const [data, setData] = useState(null); - const [licenseInfo, setLicenseInfo] = useState(null); const [isLoading, setIsLoading] = useState(false); const [error, setError] = useState(null); + const licenseInfo = useLicenseInfo(accessToken).data ?? null; + useEffect(() => { const fetchData = async () => { if (!accessToken) return; @@ -70,12 +62,8 @@ export default function UsageIndicator({ accessToken, width = 220 }: UsageIndica setError(null); try { - const [usageResult, licenseResult] = await Promise.all([ - getRemainingUsers(accessToken), - getLicenseInfo(accessToken).catch(() => null), // Don't fail if license endpoint unavailable - ]); + const usageResult = await getRemainingUsers(accessToken); setData(usageResult); - setLicenseInfo(licenseResult); } catch (err) { console.error("Failed to fetch usage data:", err); setError("Failed to load usage data"); diff --git a/ui/litellm-dashboard/src/components/UsagePage/components/EndpointUsage/components/EndpointUsageTable.test.tsx b/ui/litellm-dashboard/src/components/UsagePage/components/EndpointUsage/components/EndpointUsageTable.test.tsx index 793e4c6e3cf..63c606eb00e 100644 --- a/ui/litellm-dashboard/src/components/UsagePage/components/EndpointUsage/components/EndpointUsageTable.test.tsx +++ b/ui/litellm-dashboard/src/components/UsagePage/components/EndpointUsage/components/EndpointUsageTable.test.tsx @@ -31,12 +31,6 @@ vi.mock("antd", async () => { return { Table, Progress }; }); -vi.mock("@/utils/dataUtils", () => ({ - formatNumberWithCommas: (value: number, decimals?: number) => { - return value.toFixed(decimals || 0); - }, -})); - describe("EndpointUsageTable", () => { it("should render", () => { const mockEndpointData = { diff --git a/ui/litellm-dashboard/src/components/UsagePage/components/EndpointUsage/components/EndpointUsageTable.tsx b/ui/litellm-dashboard/src/components/UsagePage/components/EndpointUsage/components/EndpointUsageTable.tsx index f5cfb553370..88f92360ab5 100644 --- a/ui/litellm-dashboard/src/components/UsagePage/components/EndpointUsage/components/EndpointUsageTable.tsx +++ b/ui/litellm-dashboard/src/components/UsagePage/components/EndpointUsage/components/EndpointUsageTable.tsx @@ -1,7 +1,7 @@ import React from "react"; import { Table, Progress } from "antd"; import type { ColumnsType } from "antd/es/table"; -import { formatNumberWithCommas } from "@/utils/dataUtils"; +import { MoneyCell } from "@/components/shared/table_cells"; import { MetricWithMetadata } from "../../../types"; interface EndpointUsageTableProps { @@ -112,7 +112,7 @@ const EndpointUsageTable: React.FC = ({ endpointData }) title: "Spend", dataIndex: "spend", key: "spend", - render: (value: number) => `$${formatNumberWithCommas(value, 2)}`, + render: (value: number) => , }, ]; diff --git a/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/EntityUsage.tsx b/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/EntityUsage.tsx index df5b14a57d6..9dd72f44f67 100644 --- a/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/EntityUsage.tsx +++ b/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/EntityUsage.tsx @@ -1,4 +1,5 @@ import useTeams from "@/app/(dashboard)/hooks/useTeams"; +import { MoneyCell } from "@/components/shared/table_cells"; import { formatNumberWithCommas } from "@/utils/dataUtils"; import { BarChart, @@ -665,7 +666,9 @@ const EntityUsage: React.FC = ({ accessToken, entityType, enti .map((entity) => ( {entity.metadata.alias} - ${formatNumberWithCommas(entity.metrics.spend, 4)} + + + {entity.metrics.successful_requests.toLocaleString()} @@ -777,7 +780,9 @@ const EntityUsage: React.FC = ({ accessToken, entityType, enti {provider.provider}
- ${formatNumberWithCommas(provider.spend, 2)} + + + {provider.successful_requests.toLocaleString()} diff --git a/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/SpendByProvider.test.tsx b/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/SpendByProvider.test.tsx index 0541a9c6925..7eea7653ef1 100644 --- a/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/SpendByProvider.test.tsx +++ b/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/SpendByProvider.test.tsx @@ -1,4 +1,4 @@ -import { render, screen } from "@testing-library/react"; +import { fireEvent, render, screen } from "@testing-library/react"; import { describe, expect, it, vi, beforeEach } from "vitest"; import SpendByProvider from "./SpendByProvider"; @@ -200,6 +200,14 @@ describe("SpendByProvider", () => { expect(screen.getByText("1,234,567")).toBeInTheDocument(); }); + it("should render zero spend as a dash when Show Zero Spend is on", () => { + render(); + fireEvent.click(screen.getAllByRole("switch")[0]); + expect(screen.getAllByText("google").length).toBeGreaterThan(0); + expect(screen.getByText("-")).toBeInTheDocument(); + expect(screen.queryByText("$0.00")).not.toBeInTheDocument(); + }); + it("should filter data correctly when both toggles are off", () => { render(); expect(screen.getAllByText("openai").length).toBeGreaterThan(0); diff --git a/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/SpendByProvider.tsx b/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/SpendByProvider.tsx index 58d673bb7ad..5cea84affb9 100644 --- a/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/SpendByProvider.tsx +++ b/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/SpendByProvider.tsx @@ -1,3 +1,4 @@ +import { MoneyCell } from "@/components/shared/table_cells"; import { formatNumberWithCommas } from "@/utils/dataUtils"; import { InfoCircleOutlined } from "@ant-design/icons"; import { @@ -109,7 +110,9 @@ const SpendByProvider: React.FC = ({ loading, isDateChangi {provider.provider} - ${formatNumberWithCommas(provider.spend, 2)} + + + {provider.successful_requests.toLocaleString()} {provider.failed_requests.toLocaleString()} {provider.tokens.toLocaleString()} diff --git a/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopKeyView.test.tsx b/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopKeyView.test.tsx index 126bf51bd36..766f027434f 100644 --- a/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopKeyView.test.tsx +++ b/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopKeyView.test.tsx @@ -171,7 +171,9 @@ describe("TopKeyView", () => { ]} />, ); - expect(screen.getByText(/sk-1234\.\.\./)).toBeInTheDocument(); + const keyId = screen.getByText("sk-1234567890abcdef"); + expect(keyId).toBeInTheDocument(); + expect(keyId).toHaveClass("truncate"); }); it("should display dash for missing key alias", () => { @@ -206,7 +208,7 @@ describe("TopKeyView", () => { expect(screen.getByText("$123.46")).toBeInTheDocument(); }); - it("should display less than 0.01 spend as <$0.01", () => { + it("should display sub-cent spend as < $0.01", () => { render( { { api_key: "key-123", key_alias: "Test Key", - spend: 0.005, + spend: 0.004, }, ]} />, ); - expect(screen.getByText("<$0.01")).toBeInTheDocument(); + expect(screen.getByText("< $0.01")).toBeInTheDocument(); }); - it("should display zero spend correctly", () => { + it("should display zero spend as a dash", () => { render( { ]} />, ); - expect(screen.getByText("$0.00")).toBeInTheDocument(); + expect(screen.getByText("-")).toBeInTheDocument(); + expect(screen.queryByText("$0.00")).not.toBeInTheDocument(); }); it("should display dash for empty tags", () => { @@ -376,7 +379,7 @@ describe("TopKeyView", () => { />, ); - const keyIdButton = screen.getByText(/key-123\.\.\./).closest("button"); + const keyIdButton = screen.getByText("key-123").closest("button"); if (keyIdButton) { await user.click(keyIdButton); } @@ -410,7 +413,7 @@ describe("TopKeyView", () => { />, ); - const keyIdButton = screen.getByText(/key-123\.\.\./).closest("button"); + const keyIdButton = screen.getByText("key-123").closest("button"); if (keyIdButton) { await user.click(keyIdButton); } @@ -447,7 +450,7 @@ describe("TopKeyView", () => { />, ); - const keyIdButton = screen.getByText(/key-123\.\.\./).closest("button"); + const keyIdButton = screen.getByText("key-123").closest("button"); if (keyIdButton) { await user.click(keyIdButton); } @@ -483,7 +486,7 @@ describe("TopKeyView", () => { />, ); - const keyIdButton = screen.getByText(/key-123\.\.\./).closest("button"); + const keyIdButton = screen.getByText("key-123").closest("button"); if (keyIdButton) { await user.click(keyIdButton); } @@ -522,7 +525,7 @@ describe("TopKeyView", () => { />, ); - const keyIdButton = screen.getByText(/key-123\.\.\./).closest("button"); + const keyIdButton = screen.getByText("key-123").closest("button"); if (keyIdButton) { await user.click(keyIdButton); } @@ -552,7 +555,7 @@ describe("TopKeyView", () => { />, ); - const keyIdButton = screen.getByText(/key-123\.\.\./).closest("button"); + const keyIdButton = screen.getByText("key-123").closest("button"); if (keyIdButton) { await user.click(keyIdButton); } diff --git a/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopKeyView.tsx b/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopKeyView.tsx index d0748583c30..2dcf98a1e5c 100644 --- a/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopKeyView.tsx +++ b/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopKeyView.tsx @@ -1,6 +1,7 @@ import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; +import { IdCell, MoneyCell } from "@/components/shared/table_cells"; import { ChevronDownIcon, ChevronUpIcon } from "@heroicons/react/outline"; -import { BarChart, Button } from "@tremor/react"; +import { BarChart } from "@tremor/react"; import { Segmented, Tooltip } from "antd"; import React, { useState } from "react"; import { formatNumberWithCommas } from "../../../../utils/dataUtils"; @@ -83,20 +84,7 @@ const TopKeyView: React.FC = ({ topKeys, teams, showTags = fals { header: "Key ID", accessorKey: "api_key", - cell: (info: any) => ( -
- - - -
- ), + cell: (info: any) => handleKeyClick(info.row.original)} />, }, { header: "Key Alias", @@ -164,10 +152,8 @@ const TopKeyView: React.FC = ({ topKeys, teams, showTags = fals const spendColumn = { header: "Spend (USD)", accessorKey: "spend", - cell: (info: any) => { - const value = info.getValue(); - return value > 0 && value < 0.01 ? "<$0.01" : `$${formatNumberWithCommas(value, 2)}`; - }, + meta: { numeric: true }, + cell: (info: any) => , }; const columns = showTags ? [...baseColumns, tagsColumn, spendColumn] : [...baseColumns, spendColumn]; @@ -247,13 +233,7 @@ const TopKeyView: React.FC = ({ topKeys, teams, showTags = fals ) : (
- <>} - getRowCanExpand={() => false} - isLoading={false} - /> +
)} diff --git a/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopModelView.test.tsx b/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopModelView.test.tsx index f6014d025ae..bbf9379f5e1 100644 --- a/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopModelView.test.tsx +++ b/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopModelView.test.tsx @@ -175,7 +175,7 @@ describe("TopModelView", () => { setTopModelsLimit={mockSetTopModelsLimit} />, ); - expect(screen.getByText("$0.00")).toBeInTheDocument(); + expect(screen.getByText("-")).toBeInTheDocument(); expect(screen.getAllByText("0").length).toBeGreaterThan(0); }); diff --git a/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopModelView.tsx b/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopModelView.tsx index c69ba42f182..8938767c02c 100644 --- a/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopModelView.tsx +++ b/ui/litellm-dashboard/src/components/UsagePage/components/EntityUsage/TopModelView.tsx @@ -1,6 +1,7 @@ import { BarChart } from "@tremor/react"; import { Segmented } from "antd"; import { useState } from "react"; +import { MoneyCell } from "@/components/shared/table_cells"; import { formatNumberWithCommas } from "../../../../utils/dataUtils"; import { DataTable } from "../../../view_logs/table"; @@ -30,24 +31,25 @@ export default function TopModelView({ topModels, topModelsLimit, setTopModelsLi { header: "Spend (USD)", accessorKey: "spend", - cell: (info: any) => { - const value = info.getValue(); - return `$${formatNumberWithCommas(value, 2)}`; - }, + meta: { numeric: true }, + cell: (info: any) => , }, { header: "Successful", accessorKey: "successful_requests", + meta: { numeric: true }, cell: (info: any) => {info.getValue()?.toLocaleString() || 0}, }, { header: "Failed", accessorKey: "failed_requests", + meta: { numeric: true }, cell: (info: any) => {info.getValue()?.toLocaleString() || 0}, }, { header: "Tokens", accessorKey: "tokens", + meta: { numeric: true }, cell: (info: any) => info.getValue()?.toLocaleString() || 0, }, ]; @@ -99,13 +101,7 @@ export default function TopModelView({ topModels, topModelsLimit, setTopModelsLi ) : (
- <>} - getRowCanExpand={() => false} - isLoading={false} - /> +
)} diff --git a/ui/litellm-dashboard/src/components/UsagePage/components/KeyModelUsageView.test.tsx b/ui/litellm-dashboard/src/components/UsagePage/components/KeyModelUsageView.test.tsx index 61968294e18..322f00a501a 100644 --- a/ui/litellm-dashboard/src/components/UsagePage/components/KeyModelUsageView.test.tsx +++ b/ui/litellm-dashboard/src/components/UsagePage/components/KeyModelUsageView.test.tsx @@ -158,7 +158,7 @@ describe("KeyModelUsageView", () => { }, ]; render(); - expect(screen.getByText("$0.00")).toBeInTheDocument(); + expect(screen.getByText("-")).toBeInTheDocument(); expect(screen.getAllByText("0").length).toBeGreaterThan(0); }); diff --git a/ui/litellm-dashboard/src/components/UsagePage/components/KeyModelUsageView.tsx b/ui/litellm-dashboard/src/components/UsagePage/components/KeyModelUsageView.tsx index ee1a49051da..ceb00e8a19f 100644 --- a/ui/litellm-dashboard/src/components/UsagePage/components/KeyModelUsageView.tsx +++ b/ui/litellm-dashboard/src/components/UsagePage/components/KeyModelUsageView.tsx @@ -1,3 +1,4 @@ +import { MoneyCell } from "@/components/shared/table_cells"; import { formatNumberWithCommas } from "@/utils/dataUtils"; import { BarChart, Card, Title } from "@tremor/react"; import { Table } from "antd"; @@ -24,7 +25,7 @@ const columns: ColumnsType = [ title: "Spend (USD)", dataIndex: "spend", key: "spend", - render: (value) => `$${formatNumberWithCommas(value, 2)}`, + render: (value) => , }, { title: "Successful", diff --git a/ui/litellm-dashboard/src/components/VirtualKeysPage/VirtualKeysTable.test.tsx b/ui/litellm-dashboard/src/components/VirtualKeysPage/VirtualKeysTable.test.tsx index ba616a03fd9..02f5d588149 100644 --- a/ui/litellm-dashboard/src/components/VirtualKeysPage/VirtualKeysTable.test.tsx +++ b/ui/litellm-dashboard/src/components/VirtualKeysPage/VirtualKeysTable.test.tsx @@ -1,4 +1,5 @@ -import { act, screen, waitFor, within, fireEvent } from "@testing-library/react"; +import { screen, waitFor, within, fireEvent } from "@testing-library/react"; +import userEvent from "@testing-library/user-event"; import { vi, it, expect, beforeEach, describe, MockedFunction } from "vitest"; import { renderWithProviders } from "../../../tests/test-utils"; import { VirtualKeysTable } from "./VirtualKeysTable"; @@ -175,7 +176,7 @@ it("should display key information correctly", async () => { await waitFor(() => { expect(screen.getByText("Test Key Alias")).toBeInTheDocument(); expect(screen.getByText("Test Team")).toBeInTheDocument(); - expect(screen.getByText("5.5000")).toBeInTheDocument(); + expect(screen.getByText("$5.5000")).toBeInTheDocument(); }); }); @@ -477,9 +478,8 @@ describe("Status column reflects key.blocked / scim_blocked metadata", () => { const tag = await screen.findByTestId(`key-status-${mockKey.token_id}`); expect(tag).toHaveTextContent("Blocked"); - act(() => { - fireEvent.mouseEnter(tag); - }); + const user = userEvent.setup(); + await user.hover(tag); await waitFor(() => { expect(screen.getByText(/Blocked by SCIM/i)).toBeInTheDocument(); }); diff --git a/ui/litellm-dashboard/src/components/VirtualKeysPage/VirtualKeysTable.tsx b/ui/litellm-dashboard/src/components/VirtualKeysPage/VirtualKeysTable.tsx index a803d78ed57..cae6dc54df5 100644 --- a/ui/litellm-dashboard/src/components/VirtualKeysPage/VirtualKeysTable.tsx +++ b/ui/litellm-dashboard/src/components/VirtualKeysPage/VirtualKeysTable.tsx @@ -13,20 +13,10 @@ import { SortingState, useReactTable, } from "@tanstack/react-table"; -import { - Badge, - Button, - Icon, - Table, - TableBody, - TableCell, - TableHead, - TableHeaderCell, - TableRow, - Text, -} from "@tremor/react"; +import { Badge, Icon, Table, TableBody, TableCell, TableHead, TableHeaderCell, TableRow, Text } from "@tremor/react"; import { InfoCircleOutlined, SyncOutlined } from "@ant-design/icons"; -import { Button as AntButton, Popover, Skeleton, Tag, Tooltip, Typography } from "antd"; +import { Button as AntButton, Popover, Skeleton, Typography } from "antd"; +import { DateCell, IdCell, MoneyCell, StatusBadge } from "@/components/shared/table_cells"; import React, { useDeferredValue, useMemo, useState } from "react"; import { getModelDisplayName } from "../key_team_helpers/fetch_available_models_team_key"; import { PaginatedKeyAliasSelect } from "../KeyAliasSelect/PaginatedKeyAliasSelect/PaginatedKeyAliasSelect"; @@ -145,23 +135,7 @@ export function VirtualKeysTable() { header: "Key ID", size: 100, enableSorting: true, - cell: (info) => { - const value = info.getValue() as string; - const width = info.cell.column.getSize(); - return ( - - - - ); - }, + cell: (info) => setSelectedKey(info.row.original)} />, }, { id: "key_alias", @@ -187,22 +161,14 @@ export function VirtualKeysTable() { cell: ({ row }) => { const key = row.original; if (key.blocked !== true) { - return ( - - Active - - ); + return ; } const isScimBlocked = (key.metadata as Record | null | undefined)?.scim_blocked === true; const reason = isScimBlocked ? "Blocked by SCIM (external identity provider deactivated or deleted the owning user)." : "Blocked. Requests using this key will be rejected with 401."; return ( - - - Blocked - - + ); }, }, @@ -323,10 +289,7 @@ export function VirtualKeysTable() { header: "Created At", size: 120, enableSorting: true, - cell: (info) => { - const value = info.getValue(); - return value ? new Date(value as string).toLocaleDateString() : "-"; - }, + cell: (info) => , }, { id: "created_by", @@ -394,10 +357,7 @@ export function VirtualKeysTable() { header: "Updated At", size: 120, enableSorting: true, - cell: (info) => { - const value = info.getValue(); - return value ? new Date(value as string).toLocaleDateString() : "Never"; - }, + cell: (info) => , }, { id: "last_active", @@ -415,16 +375,7 @@ export function VirtualKeysTable() { ), size: 130, enableSorting: false, - cell: (info) => { - const value = info.getValue(); - if (!value) return "Unknown"; - const date = new Date(value as string); - return ( - - {date.toLocaleDateString()} - - ); - }, + cell: (info) => , }, { id: "expires", @@ -432,10 +383,7 @@ export function VirtualKeysTable() { header: "Expires", size: 120, enableSorting: false, - cell: (info) => { - const value = info.getValue(); - return value ? new Date(value as string).toLocaleDateString() : "Never"; - }, + cell: (info) => , }, { id: "spend", @@ -443,7 +391,7 @@ export function VirtualKeysTable() { header: "Spend (USD)", size: 100, enableSorting: true, - cell: (info) => formatNumberWithCommas(info.getValue() as number, 4), + cell: (info) => , }, { id: "max_budget", @@ -470,10 +418,7 @@ export function VirtualKeysTable() { header: "Budget Reset", size: 130, enableSorting: false, - cell: (info) => { - const value = info.getValue(); - return value ? new Date(value as string).toLocaleString() : "Never"; - }, + cell: (info) => , }, { id: "models", @@ -633,7 +578,7 @@ export function VirtualKeysTable() { }, { name: "Key Hash", - label: "Key Hash", + label: "Key ID", isSearchable: false, }, ]; diff --git a/ui/litellm-dashboard/src/components/agents.tsx b/ui/litellm-dashboard/src/components/agents.tsx index 0d8916942da..e3703f6c588 100644 --- a/ui/litellm-dashboard/src/components/agents.tsx +++ b/ui/litellm-dashboard/src/components/agents.tsx @@ -20,7 +20,7 @@ import AgentInfoView from "./agents/agent_info"; import NotificationsManager from "./molecules/notifications_manager"; import { Agent } from "./agents/types"; import { Team } from "./key_team_helpers/key_list"; -import { formatNumberWithCommas } from "@/utils/dataUtils"; +import { DateCell, IdCell, MoneyCell, StatusBadge } from "@/components/shared/table_cells"; import TableIconActionButton from "./common_components/IconActionButton/TableIconActionButtons/TableIconActionButton"; interface AgentsPanelProps { @@ -193,19 +193,10 @@ const AgentsPanel: React.FC = ({ accessToken, userRole, teams {agent.agent_name} - - - + setSelectedAgentId(id)} /> - {formatNumberWithCommas(agent.spend, 4)} + @@ -213,13 +204,13 @@ const AgentsPanel: React.FC = ({ accessToken, userRole, teams - {agent.created_at ? new Date(agent.created_at).toLocaleDateString() : "N/A"} + {(agent.keys?.length ?? 0) > 0 ? ( - Active + ) : ( - Needs Setup + )} {isAdmin && ( diff --git a/ui/litellm-dashboard/src/components/bulk_create_users_button.tsx b/ui/litellm-dashboard/src/components/bulk_create_users_button.tsx index 17690c988dc..23b69722546 100644 --- a/ui/litellm-dashboard/src/components/bulk_create_users_button.tsx +++ b/ui/litellm-dashboard/src/components/bulk_create_users_button.tsx @@ -361,7 +361,7 @@ const BulkCreateUsersButton: React.FC = ({ if (!uiSettings?.SSO_ENABLED) { // Regular invitation flow const invitationData = await invitationCreateCall(accessToken, user_id); - const invitationUrl = new URL(`/ui?invitation_id=${invitationData.id}`, baseUrl).toString(); + const invitationUrl = new URL(`/ui/onboarding?invitation_id=${invitationData.id}`, baseUrl).toString(); setParsedData((current) => current.map((u, i) => diff --git a/ui/litellm-dashboard/src/components/chat_ui/ReasoningContent.test.tsx b/ui/litellm-dashboard/src/components/chat_ui/ReasoningContent.test.tsx new file mode 100644 index 00000000000..35540d3ebde --- /dev/null +++ b/ui/litellm-dashboard/src/components/chat_ui/ReasoningContent.test.tsx @@ -0,0 +1,32 @@ +import { render, screen, fireEvent } from "@testing-library/react"; +import { describe, it, expect } from "vitest"; +import ReasoningContent from "./ReasoningContent"; + +describe("ReasoningContent", () => { + it("should render nothing when reasoningContent is empty", () => { + const { container } = render(); + expect(container).toBeEmptyDOMElement(); + }); + + it("should show reasoning content expanded by default and toggle on click", () => { + render(); + + expect(screen.getByText("thinking hard")).toBeInTheDocument(); + expect(screen.getByText("Hide reasoning")).toBeInTheDocument(); + + fireEvent.click(screen.getByRole("button")); + + expect(screen.queryByText("thinking hard")).not.toBeInTheDocument(); + expect(screen.getByText("Show reasoning")).toBeInTheDocument(); + }); + + it("should constrain width and break long words so it cannot expand the layout (regression #32481)", () => { + const longToken = "a".repeat(500); + render(); + + const contentBox = screen.getByText(longToken).closest("div.mt-2"); + expect(contentBox).not.toBeNull(); + expect(contentBox).toHaveClass("max-w-full"); + expect(contentBox).toHaveStyle({ wordBreak: "break-word", overflowWrap: "break-word" }); + }); +}); diff --git a/ui/litellm-dashboard/src/components/chat_ui/ReasoningContent.tsx b/ui/litellm-dashboard/src/components/chat_ui/ReasoningContent.tsx index e537c6de79f..30ba2d3fd95 100644 --- a/ui/litellm-dashboard/src/components/chat_ui/ReasoningContent.tsx +++ b/ui/litellm-dashboard/src/components/chat_ui/ReasoningContent.tsx @@ -27,7 +27,10 @@ const ReasoningContent: React.FC = ({ reasoningContent }) {isExpanded && ( -
+
= ({ reasoningContent }) language={match[1]} PreTag="div" className="rounded-md my-2" + wrapLines={true} + wrapLongLines={true} {...props} > {String(children).replace(/\n$/, "")} ) : ( - + {children} ); }, + pre: ({ node, ...props }) => 
,
             }}
           >
             {reasoningContent}
diff --git a/ui/litellm-dashboard/src/components/claude_code_plugins/plugin_table.tsx b/ui/litellm-dashboard/src/components/claude_code_plugins/plugin_table.tsx
index 1646161891f..0932e658899 100644
--- a/ui/litellm-dashboard/src/components/claude_code_plugins/plugin_table.tsx
+++ b/ui/litellm-dashboard/src/components/claude_code_plugins/plugin_table.tsx
@@ -11,6 +11,7 @@ import {
 import { Badge, Button, Table, TableBody, TableCell, TableHead, TableHeaderCell, TableRow } from "@tremor/react";
 import { Tooltip } from "antd";
 import React, { useState } from "react";
+import { DateCell, IdCell, StatusBadge } from "@/components/shared/table_cells";
 import NotificationsManager from "../molecules/notifications_manager";
 import { getCategoryBadgeColor } from "./helpers";
 import { Plugin } from "./types";
@@ -34,12 +35,6 @@ const PluginTable: React.FC = ({
 }) => {
   const [sorting, setSorting] = useState([{ id: "created_at", desc: true }]);
 
-  const formatDate = (dateString?: string) => {
-    if (!dateString) return "-";
-    const date = new Date(dateString);
-    return date.toLocaleString();
-  };
-
   const copyToClipboard = (text: string) => {
     navigator.clipboard.writeText(text);
     NotificationsManager.success("Copied to clipboard!");
@@ -51,19 +46,9 @@ const PluginTable: React.FC = ({
       accessorKey: "name",
       cell: ({ row }) => {
         const plugin = row.original;
-        const name = plugin.name || "";
         return (
           

-            
-              
-            
+             onPluginClick(plugin.id)} />
             
                {
@@ -122,24 +107,13 @@ const PluginTable: React.FC = ({
       accessorKey: "enabled",
       cell: ({ row }) => {
         const plugin = row.original;
-        return (
-          
-            {plugin.enabled ? "Yes" : "No"}
-          
-        );
+        return ;
       },
     },
     {
       header: "Created At",
       accessorKey: "created_at",
-      cell: ({ row }) => {
-        const plugin = row.original;
-        return (
-          
-            {formatDate(plugin.created_at)}
-          
-        );
-      },
+      cell: ({ row }) => ,
     },
     ...(isAdmin
       ? [
diff --git a/ui/litellm-dashboard/src/components/guardrails/guardrail_table.tsx b/ui/litellm-dashboard/src/components/guardrails/guardrail_table.tsx
index ecf6ce48fde..99f6b2793fd 100644
--- a/ui/litellm-dashboard/src/components/guardrails/guardrail_table.tsx
+++ b/ui/litellm-dashboard/src/components/guardrails/guardrail_table.tsx
@@ -1,8 +1,8 @@
 import React, { useState } from "react";
-import { Table, TableBody, TableCell, TableHead, TableHeaderCell, TableRow, Icon, Button } from "@tremor/react";
+import { Table, TableBody, TableCell, TableHead, TableHeaderCell, TableRow, Icon } from "@tremor/react";
 import { TrashIcon, SwitchVerticalIcon, ChevronUpIcon, ChevronDownIcon } from "@heroicons/react/outline";
 import { Tooltip } from "antd";
-import { Badge } from "@tremor/react";
+import { DateCell, IdCell, StatusBadge } from "@/components/shared/table_cells";
 import {
   ColumnDef,
   flexRender,
@@ -43,13 +43,6 @@ const GuardrailTable: React.FC = ({
   const [editModalVisible, setEditModalVisible] = useState(false);
   const [selectedGuardrail, setSelectedGuardrail] = useState(null);
 
-  // Format date helper function
-  const formatDate = (dateString?: string) => {
-    if (!dateString) return "-";
-    const date = new Date(dateString);
-    return date.toLocaleString();
-  };
-
   const handleEditClick = (guardrail: Guardrail) => {
     setSelectedGuardrail(guardrail);
     setEditModalVisible(true);
@@ -65,18 +58,7 @@ const GuardrailTable: React.FC = ({
     {
       header: "Guardrail ID",
       accessorKey: "guardrail_id",
-      cell: (info: any) => (
-        
-          
-        
-      ),
+      cell: (info: any) => ,
     },
     {
       header: "Name",
@@ -126,41 +108,21 @@ const GuardrailTable: React.FC = ({
       header: "Default On",
       accessorKey: "litellm_params.default_on",
       cell: ({ row }) => {
-        const guardrail = row.original;
+        const isDefaultOn = !!row.original.litellm_params?.default_on;
         return (
-          
-            {guardrail.litellm_params?.default_on ? "Default On" : "Default Off"}
-          
+          
         );
       },
     },
     {
       header: "Created At",
       accessorKey: "created_at",
-      cell: ({ row }) => {
-        const guardrail = row.original;
-        return (
-          
-            {formatDate(guardrail.created_at)}
-          
-        );
-      },
+      cell: ({ row }) => ,
     },
     {
       header: "Updated At",
       accessorKey: "updated_at",
-      cell: ({ row }) => {
-        const guardrail = row.original;
-        return (
-          
-            {formatDate(guardrail.updated_at)}
-          
-        );
-      },
+      cell: ({ row }) => ,
     },
     {
       id: "actions",
diff --git a/ui/litellm-dashboard/src/components/mcp_hub_table_columns.tsx b/ui/litellm-dashboard/src/components/mcp_hub_table_columns.tsx
index 7cf0d48a49f..1e25f87d262 100644
--- a/ui/litellm-dashboard/src/components/mcp_hub_table_columns.tsx
+++ b/ui/litellm-dashboard/src/components/mcp_hub_table_columns.tsx
@@ -2,6 +2,7 @@ import { ColumnDef } from "@tanstack/react-table";
 import { Button, Badge, Text } from "@tremor/react";
 import { Tooltip, Tag } from "antd";
 import { CopyOutlined, InfoCircleOutlined } from "@ant-design/icons";
+import { StatusBadge, type StatusTone } from "@/components/shared/table_cells";
 
 export interface MCPServerData {
   server_id: string;
@@ -124,21 +125,17 @@ export const mcpHubColumns = (
       cell: ({ row }) => {
         const server = row.original;
 
-        const statusColors: Record = {
-          active: "green",
-          inactive: "red",
-          unknown: "gray",
-          healthy: "green",
-          unhealthy: "red",
+        const statusTones: Record = {
+          active: "success",
+          inactive: "error",
+          unknown: "neutral",
+          healthy: "success",
+          unhealthy: "error",
         };
 
-        const color = statusColors[server.status] || "gray";
+        const tone = statusTones[server.status] || "neutral";
 
-        return (
-          
-            {server.status || "unknown"}
-          
-        );
+        return ;
       },
     },
     {
diff --git a/ui/litellm-dashboard/src/components/mcp_tools/MCPToolsetsTab.tsx b/ui/litellm-dashboard/src/components/mcp_tools/MCPToolsetsTab.tsx
index 0198b830229..5e7e99ee8ec 100644
--- a/ui/litellm-dashboard/src/components/mcp_tools/MCPToolsetsTab.tsx
+++ b/ui/litellm-dashboard/src/components/mcp_tools/MCPToolsetsTab.tsx
@@ -6,6 +6,7 @@ import { ColumnDef } from "@tanstack/react-table";
 import { useMCPToolsets } from "@/app/(dashboard)/hooks/mcpServers/useMCPToolsets";
 import { useMCPServers } from "@/app/(dashboard)/hooks/mcpServers/useMCPServers";
 import { useQueryClient } from "@tanstack/react-query";
+import { DateCell, IdCell } from "@/components/shared/table_cells";
 import { DataTable } from "../view_logs/table";
 import { createMCPToolset, updateMCPToolset, deleteMCPToolset, listMCPTools, getProxyBaseUrl } from "../networking";
 import { MCPToolset, MCPToolsetTool } from "./types";
@@ -302,11 +303,7 @@ function toolsetColumns(
     {
       header: "Toolset ID",
       accessorKey: "toolset_id",
-      cell: ({ row }) => (
-        
-          {row.original.toolset_id.slice(0, 8)}…
-        
-      ),
+      cell: ({ row }) => ,
     },
     {
       header: "Name",
@@ -359,11 +356,7 @@ function toolsetColumns(
     {
       header: "Created",
       accessorKey: "created_at",
-      cell: ({ row }) => (
-        
-          {row.original.created_at ? new Date(row.original.created_at).toLocaleDateString() : "—"}
-        
-      ),
+      cell: ({ row }) => ,
     },
     ...(isAdmin
       ? [
@@ -509,8 +502,6 @@ export function MCPToolsetsTab({ accessToken, userRole }: MCPToolsetsTabProps) {
        
}
-        getRowCanExpand={() => false}
         isLoading={isLoading}
         noDataMessage="No toolsets yet. Click 'New Toolset' to create one."
         loadingMessage="Loading toolsets..."
diff --git a/ui/litellm-dashboard/src/components/mcp_tools/PassthroughAuthorizeSection.tsx b/ui/litellm-dashboard/src/components/mcp_tools/PassthroughAuthorizeSection.tsx
new file mode 100644
index 00000000000..af81f2713ae
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/mcp_tools/PassthroughAuthorizeSection.tsx
@@ -0,0 +1,75 @@
+import React from "react";
+import { Button, Form, Input } from "antd";
+import { isClientForwardedTokenMode } from "./types";
+
+interface PassthroughOAuthFlow {
+  startOAuthFlow: () => void | Promise;
+  status: string;
+  error: string | null;
+  tokenResponse: { access_token?: string; expires_in?: number } | null;
+}
+
+/**
+ * Browser-only Authorize & Fetch for the client-forwarded token modes
+ * (true_passthrough / oauth_delegate). LiteLLM never stores upstream
+ * credentials for these modes, so the token obtained here lives in this
+ * browser session only: it is forwarded per-server for the tools preview and
+ * allowlist configuration, and is never written to the server row or the
+ * per-user credential store. The optional client credentials cover IdPs
+ * without dynamic client registration (e.g. a pre-registered Slack app) and
+ * ride the temporary authorize session only.
+ */
+export default function PassthroughAuthorizeSection({
+  authType,
+  oauthFlow,
+}: {
+  authType?: string | null;
+  oauthFlow: PassthroughOAuthFlow;
+}) {
+  if (!isClientForwardedTokenMode(authType)) return null;
+  const authorizeButtonLabels: Record = {
+    authorizing: "Waiting for authorization...",
+    exchanging: "Exchanging authorization code...",
+  };
+  const authorizeButtonLabel = authorizeButtonLabels[oauthFlow.status] ?? "Authorize & Fetch Tools (browser-only)";
+  return (
+    

+      

+        Callers bring their own upstream token for this auth type, so LiteLLM stores no upstream credentials. To preview
+        tools and configure the tool allowlist, authorize against the upstream here: the token stays in this browser
+        session only and is never saved to LiteLLM.
+      

+      OAuth Client ID (optional, not saved)}
+        name={["credentials", "client_id"]}
+        extra="Only needed when the upstream does not support dynamic client registration (e.g. a pre-registered Slack app). Used for this browser authorization only."
+      >
+        
+      
+      OAuth Client Secret (optional, not saved)}
+        name={["credentials", "client_secret"]}
+      >
+        
+      
+      
+      {oauthFlow.error && 
{oauthFlow.error}
}
+      {oauthFlow.status === "success" && oauthFlow.tokenResponse?.access_token && (
+        

+          Token held for this browser session. Tools can now be previewed and configured; nothing was saved to LiteLLM.
+        

+      )}
+    

+  );
+}
diff --git a/ui/litellm-dashboard/src/components/mcp_tools/TruePassthroughWarning.tsx b/ui/litellm-dashboard/src/components/mcp_tools/TruePassthroughWarning.tsx
new file mode 100644
index 00000000000..b52d3f4c672
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/mcp_tools/TruePassthroughWarning.tsx
@@ -0,0 +1,21 @@
+import React from "react";
+import { Alert } from "antd";
+import { AUTH_TYPE } from "./types";
+
+/**
+ * Warning shown in the create/edit MCP server forms when auth_type
+ * true_passthrough is selected: the gateway performs no admission auth for
+ * that server, so callers reach the upstream without a LiteLLM identity.
+ */
+export default function TruePassthroughWarning({ authType }: { authType?: string | null }) {
+  if (authType !== AUTH_TYPE.TRUE_PASSTHROUGH) return null;
+  return (
+    
+  );
+}
diff --git a/ui/litellm-dashboard/src/components/mcp_tools/create_mcp_server.test.tsx b/ui/litellm-dashboard/src/components/mcp_tools/create_mcp_server.test.tsx
index 36af2f8d9fc..02374a3ffa4 100644
--- a/ui/litellm-dashboard/src/components/mcp_tools/create_mcp_server.test.tsx
+++ b/ui/litellm-dashboard/src/components/mcp_tools/create_mcp_server.test.tsx
@@ -30,6 +30,7 @@ const oauthHook = vi.hoisted(() => ({
   onTokenReceived: null as
     | ((token: Record | null, registeredClient?: { clientId?: string; clientSecret?: string }) => void)
     | null,
+  getCredentials: null as (() => Record | undefined) | null,
 }));
 vi.mock("@/hooks/useMcpOAuthFlow", () => ({
   useMcpOAuthFlow: (opts: {
@@ -37,8 +38,10 @@ vi.mock("@/hooks/useMcpOAuthFlow", () => ({
       token: Record | null,
       registeredClient?: { clientId?: string; clientSecret?: string },
     ) => void;
+    getCredentials?: () => Record | undefined;
   }) => {
     oauthHook.onTokenReceived = opts.onTokenReceived;
+    oauthHook.getCredentials = opts.getCredentials ?? null;
     return {
       startOAuthFlow: vi.fn(),
       status: "idle",
@@ -164,6 +167,57 @@ describe("CreateMCPServer", () => {
       });
     });
 
+    it("should warn that LiteLLM auth is disabled when True Passthrough is selected", async () => {
+      await selectHttpTransport();
+
+      await selectAntOption("Authentication", "True Passthrough (no LiteLLM auth)");
+
+      await waitFor(() => {
+        expect(
+          screen.getByText("True Passthrough disables LiteLLM authentication for this server"),
+        ).toBeInTheDocument();
+      });
+    });
+
+    it("should not show the True Passthrough warning when OAuth Delegate is selected", async () => {
+      await selectHttpTransport();
+
+      await selectAntOption("Authentication", "OAuth Delegate (client-supplied upstream token)");
+
+      await waitFor(() => {
+        expect(screen.getAllByText("OAuth Delegate (client-supplied upstream token)").length).toBeGreaterThan(0);
+      });
+      expect(
+        screen.queryByText("True Passthrough disables LiteLLM authentication for this server"),
+      ).not.toBeInTheDocument();
+    });
+
+    it.each([["True Passthrough (no LiteLLM auth)"], ["OAuth Delegate (client-supplied upstream token)"]])(
+      "should show the browser-only authorize section when %s is selected",
+      async (optionLabel) => {
+        await selectHttpTransport();
+
+        await selectAntOption("Authentication", optionLabel);
+
+        await waitFor(() => {
+          expect(screen.getByRole("button", { name: "Authorize & Fetch Tools (browser-only)" })).toBeInTheDocument();
+        });
+        expect(screen.getByText("OAuth Client ID (optional, not saved)")).toBeInTheDocument();
+        expect(screen.getByText("OAuth Client Secret (optional, not saved)")).toBeInTheDocument();
+      },
+    );
+
+    it("should not show the browser-only authorize section for API Key auth", async () => {
+      await selectHttpTransport();
+
+      await selectAntOption("Authentication", "API Key");
+
+      await waitFor(() => {
+        expect(screen.getByText("Authentication Value")).toBeInTheDocument();
+      });
+      expect(screen.queryByRole("button", { name: "Authorize & Fetch Tools (browser-only)" })).not.toBeInTheDocument();
+    });
+
     it("should not require auth value when creating a server with API Key auth type", async () => {
       await selectHttpTransport();
 
@@ -298,6 +352,90 @@ describe("CreateMCPServer", () => {
       expect(payload.credentials).toEqual({ auth_value: "my-secret-key" });
     });
 
+    it("does not write the browser-authorized token into form.credentials for true_passthrough", async () => {
+      await selectHttpTransport();
+
+      const user = userEvent.setup({ delay: null });
+      await user.type(getServerNameInput(), "PT_Server");
+      await user.type(screen.getByPlaceholderText("https://your-mcp-server.com"), "https://example.com/mcp");
+
+      await selectAntOption("Authentication", "True Passthrough (no LiteLLM auth)");
+
+      // Simulate the browser Authorize & Fetch flow handing back an upstream token.
+      await waitFor(() => expect(oauthHook.onTokenReceived).toBeTruthy());
+      await act(async () => {
+        oauthHook.onTokenReceived!({ access_token: "upstream-tok", token_type: "Bearer" }, undefined);
+      });
+
+      // For a browser-only mode the token must never land in form.credentials, which the OAuth flow's
+      // getCredentials reads for preview requests and the redirect-persist cache serializes. Without
+      // the guard, onTokenReceived writes it here and this returns { access_token: "upstream-tok" }.
+      const credentials = oauthHook.getCredentials?.() ?? {};
+      expect(credentials.access_token).toBeUndefined();
+    });
+
+    it.each([
+      ["true_passthrough", "True Passthrough (no LiteLLM auth)"],
+      ["oauth_delegate", "OAuth Delegate (client-supplied upstream token)"],
+    ])("persists only tool config on create for %s; the token stays browser-held", async (_authType, optionLabel) => {
+      oauthHook.tokenResponse = { access_token: "upstream-tok", token_type: "Bearer" };
+      await selectHttpTransport();
+
+      const user = userEvent.setup({ delay: null });
+      await user.type(getServerNameInput(), "CF_Server");
+      await user.type(screen.getByPlaceholderText("https://your-mcp-server.com"), "https://example.com/mcp");
+
+      await selectAntOption("Authentication", optionLabel);
+
+      await waitFor(() => expect(oauthHook.onTokenReceived).toBeTruthy());
+      await act(async () => {
+        oauthHook.onTokenReceived!({ access_token: "upstream-tok", token_type: "Bearer" }, undefined);
+      });
+
+      fireEvent.click(screen.getByRole("button", { name: "Disable all tools" }));
+
+      // Previewing and configuring must stay stateless: nothing is persisted anywhere (server row,
+      // per-user DB credential, sessionStorage) until the admin submits.
+      expect(networking.createMCPServer).not.toHaveBeenCalled();
+      expect(networking.storeMCPOAuthUserCredential).not.toHaveBeenCalled();
+      expect(setToken).not.toHaveBeenCalled();
+
+      const createdServer = {
+        server_id: "new-cf-server",
+        server_name: "CF_Server",
+        alias: "CF_Server",
+        url: "https://example.com/mcp",
+        transport: "http",
+        auth_type: _authType,
+        created_at: "2024-01-01T00:00:00Z",
+        created_by: "user-1",
+        updated_at: "2024-01-01T00:00:00Z",
+        updated_by: "user-1",
+      };
+      vi.mocked(networking.createMCPServer).mockResolvedValue(createdServer);
+
+      const submitButton = screen.getByRole("button", { name: "Add MCP Server" });
+      await act(async () => {
+        fireEvent.click(submitButton);
+      });
+
+      await waitFor(() => expect(networking.createMCPServer).toHaveBeenCalledTimes(1));
+      const [, payload] = vi.mocked(networking.createMCPServer).mock.calls[0];
+
+      // Only the tool configuration persists on the server row; the upstream token appears nowhere
+      // in the create payload and no per-user DB credential is written. The token is committed to
+      // sessionStorage only, keyed to the created server.
+      expect(payload.allowed_tools).toEqual([]);
+      expect(payload.credentials).toBeUndefined();
+      expect(JSON.stringify(payload)).not.toContain("upstream-tok");
+      expect(networking.storeMCPOAuthUserCredential).not.toHaveBeenCalled();
+      expect(setToken).toHaveBeenCalledWith(
+        "new-cf-server",
+        expect.objectContaining({ access_token: "upstream-tok" }),
+        undefined,
+      );
+    });
+
     it("should not show auth value field when None auth type is selected", async () => {
       await selectHttpTransport();
 
@@ -605,6 +743,93 @@ describe("CreateMCPServer", () => {
       // Asserted in setupOAuthInteractive
     });
 
+    it("invalidates the held token when the auth mode changes after Authorize & Fetch", async () => {
+      await setupOAuthInteractive();
+      const urlInput = screen.getByPlaceholderText("https://your-mcp-server.com");
+      await act(async () => {
+        fireEvent.change(urlInput, { target: { value: "https://a.example.com/mcp" } });
+      });
+      act(() => {
+        oauthHook.onTokenReceived?.({ access_token: "tok-a" }, { clientId: "client-a", clientSecret: "secret-a" });
+      });
+      oauthHook.reset.mockClear();
+
+      // Switching the Authentication mode changes the OAuth identity, so the held token is discarded.
+      await selectAntOption("Authentication", "True Passthrough (no LiteLLM auth)");
+
+      await waitFor(() => expect(oauthHook.reset).toHaveBeenCalled());
+    });
+
+    it("does NOT invalidate the held token when a non-mint field (server name) changes", async () => {
+      await setupOAuthInteractive();
+      const urlInput = screen.getByPlaceholderText("https://your-mcp-server.com");
+      await act(async () => {
+        fireEvent.change(urlInput, { target: { value: "https://a.example.com/mcp" } });
+      });
+      act(() => {
+        oauthHook.onTokenReceived?.({ access_token: "tok-a" }, { clientId: "client-a", clientSecret: "secret-a" });
+      });
+      oauthHook.reset.mockClear();
+
+      const nameInput = document.getElementById("server_name") as HTMLInputElement;
+      await act(async () => {
+        fireEvent.change(nameInput, { target: { value: "Renamed_Server" } });
+      });
+
+      // server_name is not part of the OAuth identity, so the held token must survive the edit.
+      await waitFor(() => expect(screen.getAllByRole("button", { name: "Add MCP Server" }).length).toBeGreaterThan(0));
+      expect(oauthHook.reset).not.toHaveBeenCalled();
+    });
+
+    it("does not refetch the tool preview with a discarded token after invalidation", async () => {
+      // Regression: handleFormValuesChange used to publish the pre-reset antd snapshot into
+      // formValues after clearHeldOAuthToken, so useTestMCPConnection kept the discarded OAuth
+      // material (the DCR client minted for the old identity) and sent it on the next tool-preview
+      // request.
+      await setupOAuthInteractive();
+      const urlInput = screen.getByPlaceholderText("https://your-mcp-server.com");
+      await act(async () => {
+        fireEvent.change(urlInput, { target: { value: "https://a.example.com/mcp" } });
+      });
+      act(() => {
+        oauthHook.onTokenReceived?.({ access_token: "stale-tok" }, { clientId: "client-a", clientSecret: "secret-a" });
+      });
+      const nameInput = document.getElementById("server_name") as HTMLInputElement;
+      await act(async () => {
+        fireEvent.change(nameInput, { target: { value: "Sync_FormValues" } });
+      });
+      vi.mocked(networking.testMCPToolsListRequest).mockClear();
+
+      await selectAntOption("Authentication", "API Key");
+
+      await waitFor(() => expect(vi.mocked(networking.testMCPToolsListRequest)).toHaveBeenCalled());
+      for (const call of vi.mocked(networking.testMCPToolsListRequest).mock.calls) {
+        expect(call[1]?.credentials?.client_id).not.toBe("client-a");
+        expect(call[1]?.credentials?.client_secret).not.toBe("secret-a");
+      }
+    });
+
+    it("keeps the held token on an http to sse switch with the same url", async () => {
+      // Same url means the same resource/audience (RFC 8707): the minted token is still valid, so a
+      // pure transport swap between the two MCP wire protocols must not force a re-authorize.
+      await setupOAuthInteractive();
+      const urlInput = screen.getByPlaceholderText("https://your-mcp-server.com");
+      await act(async () => {
+        fireEvent.change(urlInput, { target: { value: "https://a.example.com/mcp" } });
+      });
+      act(() => {
+        oauthHook.onTokenReceived?.({ access_token: "tok-a" }, { clientId: "client-a", clientSecret: "secret-a" });
+      });
+      oauthHook.reset.mockClear();
+
+      await selectAntOption("Transport Type", "Server-Sent Events (SSE)");
+
+      await waitFor(() => {
+        expect(screen.getByPlaceholderText("https://your-mcp-server.com")).toBeInTheDocument();
+      });
+      expect(oauthHook.reset).not.toHaveBeenCalled();
+    });
+
     it("includes token_validation in payload when token_validation_json is filled with valid JSON", async () => {
       vi.mocked(networking.createMCPServer).mockResolvedValue({
         server_id: "new-server-oauth",
diff --git a/ui/litellm-dashboard/src/components/mcp_tools/create_mcp_server.tsx b/ui/litellm-dashboard/src/components/mcp_tools/create_mcp_server.tsx
index 10668468c15..eb48fd02474 100644
--- a/ui/litellm-dashboard/src/components/mcp_tools/create_mcp_server.tsx
+++ b/ui/litellm-dashboard/src/components/mcp_tools/create_mcp_server.tsx
@@ -14,8 +14,14 @@ import {
   getMcpOAuthMode,
   MCP_OAUTH2_FLOW_M2M,
   MCP_OAUTH2_FLOW_INTERACTIVE,
+  isClientForwardedTokenMode,
+  getOAuthAuthorizationIdentity,
+  CLEARED_ON_INVALIDATION,
+  isHeldOAuthTokenStale,
 } from "./types";
 import OAuthFormFields from "./OAuthFormFields";
+import TruePassthroughWarning from "./TruePassthroughWarning";
+import PassthroughAuthorizeSection from "./PassthroughAuthorizeSection";
 import TokenExchangeFormFields from "./TokenExchangeFormFields";
 import MCPServerCostConfig from "./mcp_server_cost_config";
 import MCPConnectionStatus from "./mcp_connection_status";
@@ -96,7 +102,10 @@ const CreateMCPServer: React.FC = ({
   const [oauthAccessToken, setOauthAccessToken] = useState(null);
   const [logoUrl, setLogoUrl] = useState(undefined);
   const [oauthDocsUrl, setOauthDocsUrl] = useState(null);
-  const [authorizedUrl, setAuthorizedUrl] = useState(undefined);
+  // The OAuth authorization identity (see getOAuthAuthorizationIdentity) captured at the moment a token
+  // was fetched; undefined when no valid token is held. If any mint-relevant field diverges from this,
+  // the held token is stale and is discarded so the admin must re-authorize.
+  const [authorizedIdentity, setAuthorizedIdentity] = useState(undefined);
 
   // Single hook call shared by MCPConnectionStatus and MCPToolConfiguration to avoid duplicate requests.
   const {
@@ -122,32 +131,24 @@ const CreateMCPServer: React.FC = ({
   const isAwsSigV4AuthType = authType === AUTH_TYPE.AWS_SIGV4;
   const isM2MFlow = isOAuthAuthType && formValues.oauth_flow_type === OAUTH_FLOW.M2M;
 
-  const getOAuthAuthorizationTarget = (values: Record): string | undefined => {
-    const transport = values.transport || transportType;
-    const target = transport === TRANSPORT.OPENAPI ? values.spec_path : values.url;
-    return typeof target === "string" ? target : undefined;
-  };
-
   const persistCreateUiState = () => {
     if (typeof window === "undefined") {
       return;
     }
     try {
       const values = form.getFieldsValue(true);
-      setSecureItem(
-        CREATE_OAUTH_UI_STATE_KEY,
-        JSON.stringify({
-          modalVisible: isModalVisible,
-          formValues: values,
-          transportType,
-          costConfig,
-          allowedTools,
-          hasToolAllowlistInteraction,
-          searchValue,
-          aliasManuallyEdited,
-          logoUrl,
-        }),
-      );
+      const uiState = {
+        modalVisible: isModalVisible,
+        formValues: values,
+        transportType,
+        costConfig,
+        allowedTools,
+        hasToolAllowlistInteraction,
+        searchValue,
+        aliasManuallyEdited,
+        logoUrl,
+      };
+      setSecureItem(CREATE_OAUTH_UI_STATE_KEY, JSON.stringify(uiState));
     } catch (err) {
       console.warn("Failed to persist MCP create state", err);
     }
@@ -182,7 +183,7 @@ const CreateMCPServer: React.FC = ({
         description: values.description,
         url,
         transport: transport === TRANSPORT.OPENAPI ? "http" : transport,
-        auth_type: AUTH_TYPE.OAUTH2,
+        auth_type: isClientForwardedTokenMode(values.auth_type) ? values.auth_type : AUTH_TYPE.OAUTH2,
         credentials: values.credentials,
         authorization_url: values.authorization_url,
         token_url: values.token_url,
@@ -197,35 +198,61 @@ const CreateMCPServer: React.FC = ({
     onTokenReceived: (token, registeredClient) => {
       setOauthAccessToken(token?.access_token ?? null);
 
-      if (token?.access_token) {
-        const credentials = {
-          access_token: token.access_token,
-          ...(token.refresh_token && { refresh_token: token.refresh_token }),
-          ...(token.expires_in && { expires_in: token.expires_in }),
-          ...(token.scope && { scope: token.scope }),
-          ...(registeredClient?.clientId && { client_id: registeredClient.clientId }),
-          ...(registeredClient?.clientSecret && { client_secret: registeredClient.clientSecret }),
-        };
-
-        form.setFieldsValue({ credentials });
-        setAuthorizedUrl(getOAuthAuthorizationTarget(form.getFieldsValue(true)));
-
-        NotificationsManager.success(
-          "OAuth authorization successful! Please click 'Create MCP Server' to save the configuration.",
-        );
+      if (!token?.access_token) {
+        return;
       }
+
+      if (isClientForwardedTokenMode(form.getFieldValue("auth_type"))) {
+        // Browser-only modes: the token is held in local state (oauthAccessToken) for tool preview
+        // and committed to sessionStorage on submit; it must never be written into form.credentials,
+        // which would persist it as server-level credentials on the created server row. Mirrors the
+        // edit form's onTokenReceived early return.
+        setAuthorizedIdentity(getOAuthAuthorizationIdentity(form.getFieldsValue(true)));
+        NotificationsManager.success(
+          "Token held for this browser session. Tools can now be previewed and configured; nothing will be saved to LiteLLM.",
+        );
+        return;
+      }
+
+      const credentials = {
+        access_token: token.access_token,
+        ...(token.refresh_token && { refresh_token: token.refresh_token }),
+        ...(token.expires_in && { expires_in: token.expires_in }),
+        ...(token.scope && { scope: token.scope }),
+        ...(registeredClient?.clientId && { client_id: registeredClient.clientId }),
+        ...(registeredClient?.clientSecret && { client_secret: registeredClient.clientSecret }),
+      };
+
+      form.setFieldsValue({ credentials });
+      // Capture the identity AFTER writing the DCR'd credentials so the held token is not spuriously
+      // invalidated by its own credential write.
+      setAuthorizedIdentity(getOAuthAuthorizationIdentity(form.getFieldsValue(true)));
+
+      NotificationsManager.success(
+        "OAuth authorization successful! Please click 'Create MCP Server' to save the configuration.",
+      );
     },
     onBeforeRedirect: persistCreateUiState,
     flowSource: "create",
   });
 
-  const clearAuthorizedOAuthState = (values: Record) => {
-    form.resetFields(["credentials", "authorization_url", "token_url", "registration_url"]);
-    form.setFieldsValue(values);
+  // Discard the held browser-authorized token and its tool preview when the authorization identity
+  // changes (or the modal closes). The CLEARED_ON_INVALIDATION form fields (shared with the edit form
+  // via types.tsx) are reset too; whatever the admin just changed (passed via changedValues) is
+  // re-applied so the invalidation never wipes their in-flight edit. Admin-typed endpoint fields are
+  // left alone (see CLEARED_ON_INVALIDATION).
+  const clearHeldOAuthToken = (changedValues: Record = {}) => {
     setOauthAccessToken(null);
     clearTools();
     resetOAuthFlow();
-    setAuthorizedUrl(undefined);
+    setAuthorizedIdentity(undefined);
+    form.resetFields([...CLEARED_ON_INVALIDATION]);
+    const preserved = Object.fromEntries(
+      CLEARED_ON_INVALIDATION.filter((key) => key in changedValues).map((key) => [key, changedValues[key]]),
+    );
+    if (Object.keys(preserved).length > 0) {
+      form.setFieldsValue(preserved);
+    }
   };
 
   React.useEffect(() => {
@@ -494,23 +521,21 @@ const CreateMCPServer: React.FC = ({
           });
           if (oauthMode === "authorization_code") {
             const scope = oauthTokenResponse.scope;
-            await storeMCPOAuthUserCredential(accessToken, response.server_id, {
+            const oauthCredentialPayload = {
               access_token: oauthTokenResponse.access_token,
               refresh_token: oauthTokenResponse.refresh_token,
               expires_in: oauthTokenResponse.expires_in,
               scopes: typeof scope === "string" && scope ? scope.split(" ") : undefined,
-            });
+            };
+            await storeMCPOAuthUserCredential(accessToken, response.server_id, oauthCredentialPayload);
           } else {
-            setToken(
-              response.server_id,
-              {
-                access_token: oauthTokenResponse.access_token,
-                expires_in: oauthTokenResponse.expires_in,
-                refresh_token: oauthTokenResponse.refresh_token,
-                token_type: oauthTokenResponse.token_type,
-              },
-              userID,
-            );
+            const browserHeldToken = {
+              access_token: oauthTokenResponse.access_token,
+              expires_in: oauthTokenResponse.expires_in,
+              refresh_token: oauthTokenResponse.refresh_token,
+              token_type: oauthTokenResponse.token_type,
+            };
+            setToken(response.server_id, browserHeldToken, userID);
           }
         }
 
@@ -564,22 +589,11 @@ const CreateMCPServer: React.FC = ({
           ? { url: undefined, command: undefined, args: undefined, env: undefined }
           : { spec_path: undefined, command: undefined, args: undefined, env: undefined };
 
-    const nextValues =
-      authorizedUrl === undefined
-        ? transportValues
-        : {
-            ...transportValues,
-            credentials: undefined,
-            authorization_url: undefined,
-            token_url: undefined,
-            registration_url: undefined,
-          };
-
-    if (authorizedUrl !== undefined) {
-      clearAuthorizedOAuthState(nextValues);
-    } else {
-      form.setFieldsValue(nextValues);
+    form.setFieldsValue(transportValues);
+    if (isHeldOAuthTokenStale(form.getFieldsValue(true), authorizedIdentity)) {
+      clearHeldOAuthToken();
     }
+    setFormValues(form.getFieldsValue(true));
   };
 
   // Generate options with existing groups and potential new group
@@ -640,27 +654,21 @@ const CreateMCPServer: React.FC = ({
       setOauthAccessToken(null);
       clearTools();
       resetOAuthFlow();
-      setAuthorizedUrl(undefined);
+      setAuthorizedIdentity(undefined);
     }
   }, [isModalVisible, form, clearTools, resetOAuthFlow]);
 
   const isAdmin = isAdminRole(userRole);
 
   const handleFormValuesChange = (changedValues: Record, allValues: Record) => {
-    const changedAuthorizationTarget = "url" in changedValues || "spec_path" in changedValues;
-    if (
-      changedAuthorizationTarget &&
-      authorizedUrl !== undefined &&
-      getOAuthAuthorizationTarget(allValues) !== authorizedUrl
-    ) {
-      const invalidated = {
-        credentials: undefined,
-        authorization_url: changedValues.authorization_url,
-        token_url: changedValues.token_url,
-        registration_url: changedValues.registration_url,
-      };
-      clearAuthorizedOAuthState(invalidated);
-      setFormValues({ ...allValues, ...invalidated });
+    // Any change to a mint-relevant field (url, auth_type, oauth_flow_type, client creds/scopes, or the
+    // authorization/token/registration endpoints — see getOAuthAuthorizationIdentity) makes a held token
+    // stale, so discard it and force a fresh authorize. When that happens, formValues must be rebuilt
+    // from the form's post-reset state, not the pre-reset allValues snapshot: the snapshot still holds
+    // the discarded token in credentials, and useTestMCPConnection reads formValues for tool preview.
+    if (isHeldOAuthTokenStale(allValues, authorizedIdentity)) {
+      clearHeldOAuthToken(changedValues);
+      setFormValues({ ...form.getFieldsValue(true), ...changedValues });
       return;
     }
     setFormValues(allValues);
@@ -970,9 +978,25 @@ const CreateMCPServer: React.FC = ({
                             OAuth
                             OAuth Token Exchange (OBO)
                             AWS SigV4 (Bedrock AgentCore MCPs)
+                            True Passthrough (no LiteLLM auth)
+                            
+                              OAuth Delegate (client-supplied upstream token)
+                            
                           
                         
 
+                        
+
+                        
+
                         {shouldShowAuthValueField && (
                            ({
   updateMCPServer: vi.fn(),
   listMCPTools: vi.fn().mockResolvedValue({ tools: [], error: null }),
   storeMCPOAuthUserCredential: vi.fn().mockResolvedValue({}),
+  testMCPToolsListRequest: vi.fn().mockResolvedValue({ tools: [], error: null }),
 }));
 
 vi.mock("../molecules/notifications_manager", () => ({
@@ -19,14 +20,27 @@ vi.mock("../molecules/notifications_manager", () => ({
   },
 }));
 
-const mockOauth: { tokenResponse: any } = { tokenResponse: null };
+const mockOauth: {
+  tokenResponse: any;
+  getTemporaryPayload: (() => Record | null) | null;
+  onTokenReceived: ((token: Record | null) => void) | null;
+  reset: ReturnType;
+} = { tokenResponse: null, getTemporaryPayload: null, onTokenReceived: null, reset: vi.fn() };
 vi.mock("@/hooks/useMcpOAuthFlow", () => ({
-  useMcpOAuthFlow: () => ({
-    startOAuthFlow: vi.fn(),
-    status: "idle",
-    error: null,
-    tokenResponse: mockOauth.tokenResponse,
-  }),
+  useMcpOAuthFlow: (opts: {
+    getTemporaryPayload?: () => Record | null;
+    onTokenReceived?: (token: Record | null) => void;
+  }) => {
+    mockOauth.getTemporaryPayload = opts?.getTemporaryPayload ?? null;
+    mockOauth.onTokenReceived = opts?.onTokenReceived ?? null;
+    return {
+      startOAuthFlow: vi.fn(),
+      status: "idle",
+      error: null,
+      tokenResponse: mockOauth.tokenResponse,
+      reset: mockOauth.reset,
+    };
+  },
 }));
 
 vi.mock("./mcp_server_cost_config", () => ({
@@ -86,10 +100,12 @@ vi.mock("./mcp_tool_configuration", () => ({
 const mockGetToken = vi.fn();
 const mockIsTokenValid = vi.fn();
 const mockSetToken = vi.fn();
+const mockRemoveToken = vi.fn();
 vi.mock("@/utils/mcpTokenStore", () => ({
   getToken: (...args: any[]) => mockGetToken(...args),
   isTokenValid: (...args: any[]) => mockIsTokenValid(...args),
   setToken: (...args: any[]) => mockSetToken(...args),
+  removeToken: (...args: unknown[]) => mockRemoveToken(...args),
 }));
 
 // ── fixtures ──────────────────────────────────────────────────────────────────
@@ -307,6 +323,69 @@ describe("MCPServerEdit (delegate auth)", () => {
   });
 });
 
+describe("MCPServerEdit (true passthrough warning)", () => {
+  beforeEach(() => {
+    vi.clearAllMocks();
+  });
+
+  const renderWithAuthType = (authType: string) =>
+    render(
+      ,
+    );
+
+  it("warns that LiteLLM auth is disabled for a true_passthrough server", async () => {
+    renderWithAuthType("true_passthrough");
+
+    await waitFor(() => {
+      expect(screen.getByText("True Passthrough disables LiteLLM authentication for this server")).toBeInTheDocument();
+    });
+  });
+
+  it("does not warn for an oauth_delegate server", async () => {
+    renderWithAuthType("oauth_delegate");
+
+    await waitFor(() => {
+      expect(screen.getAllByRole("button", { name: "Save Changes" }).length).toBeGreaterThan(0);
+    });
+    expect(
+      screen.queryByText("True Passthrough disables LiteLLM authentication for this server"),
+    ).not.toBeInTheDocument();
+  });
+
+  it("browser-authorize temp payload uses the selected auth_type, not the stored one", async () => {
+    // Stored server is oauth2; the admin switches the dropdown to true_passthrough before saving.
+    // The temp OAuth-relay payload must reflect the selection so the exchange is treated as
+    // browser-held (no DB persistence), matching onTokenReceived and the create form.
+    render(
+      ,
+    );
+
+    await selectAntOption("Authentication", "True Passthrough (no LiteLLM auth)");
+
+    await waitFor(() => {
+      expect(mockOauth.getTemporaryPayload).toBeTruthy();
+    });
+    const payload = mockOauth.getTemporaryPayload!();
+    expect(payload).toBeTruthy();
+    expect(payload?.auth_type).toBe("true_passthrough");
+  });
+});
+
 describe("MCPServerEdit (auth type switch)", () => {
   beforeEach(() => {
     vi.clearAllMocks();
@@ -382,6 +461,159 @@ describe("MCPServerEdit (auth type switch)", () => {
   });
 });
 
+describe("MCPServerEdit OAuth token invalidation", () => {
+  beforeEach(() => {
+    vi.clearAllMocks();
+  });
+
+  const renderOAuthEdit = () =>
+    render(
+      ,
+    );
+
+  it("invalidates a session-authorized token when the transport switches to stdio", async () => {
+    // Switching to stdio clears url/auth_type via programmatic form.setFieldsValue, which antd does
+    // not report through onValuesChange; the explicit recheck in handleTransportChange must catch it.
+    // Regression: the token used to survive this switch (sessionStorage + hook state kept the old
+    // token minted for the http url).
+    renderOAuthEdit();
+
+    act(() => {
+      mockOauth.onTokenReceived?.({ access_token: "tok-1" });
+    });
+    mockOauth.reset.mockClear();
+
+    await selectAntOption("Transport Type", "Standard Input/Output (stdio)");
+
+    await waitFor(() => expect(mockOauth.reset).toHaveBeenCalled());
+    expect(mockRemoveToken).toHaveBeenCalledWith("oauth_server_1", undefined);
+  });
+
+  it("invalidates a session-authorized token when the server URL changes", async () => {
+    renderOAuthEdit();
+
+    act(() => {
+      mockOauth.onTokenReceived?.({ access_token: "tok-1" });
+    });
+    mockOauth.reset.mockClear();
+
+    const urlInput = screen.getByPlaceholderText("https://your-mcp-server.com");
+    await act(async () => {
+      fireEvent.change(urlInput, { target: { value: "https://other.example.com/mcp" } });
+    });
+
+    await waitFor(() => expect(mockOauth.reset).toHaveBeenCalled());
+    expect(mockRemoveToken).toHaveBeenCalledWith("oauth_server_1", undefined);
+  });
+
+  it("previews tools with a staged interactive OAuth token before it is saved", async () => {
+    // Regression: for authorization_code the fetch went by server_id only, relying on the stored DB
+    // credential, so a token authorized in this edit session gave an empty preview until the admin
+    // saved; the create form previews the identical state via the config-based preview endpoint.
+    mockOauth.tokenResponse = { access_token: "staged-obo-tok" };
+
+    renderOAuthEdit();
+
+    await waitFor(() => {
+      expect(vi.mocked(networking.testMCPToolsListRequest)).toHaveBeenCalledWith(
+        "access-token",
+        // oauth2_flow must be explicit: the preview endpoint infers client_credentials from
+        // inherited client_id/client_secret/token_url and would strip the staged bearer.
+        expect.objectContaining({
+          server_id: "oauth_server_1",
+          url: "https://example.com/mcp",
+          oauth2_flow: "authorization_code",
+        }),
+        "staged-obo-tok",
+      );
+    });
+    expect(networking.listMCPTools).not.toHaveBeenCalled();
+    // Previewing must stay stateless: the staged token is committed only by an explicit Save
+    // (storeMCPOAuthUserCredential for authorization_code, setToken for the client-forwarded modes).
+    expect(networking.storeMCPOAuthUserCredential).not.toHaveBeenCalled();
+    expect(mockSetToken).not.toHaveBeenCalled();
+    expect(networking.updateMCPServer).not.toHaveBeenCalled();
+    mockOauth.tokenResponse = null;
+  });
+
+  it("previews an OpenAPI server's staged token against its spec_path", async () => {
+    mockOauth.tokenResponse = { access_token: "staged-obo-tok" };
+
+    render(
+      ,
+    );
+
+    await waitFor(() => {
+      expect(vi.mocked(networking.testMCPToolsListRequest)).toHaveBeenCalledWith(
+        "access-token",
+        expect.objectContaining({ spec_path: "https://example.com/openapi.json" }),
+        "staged-obo-tok",
+      );
+    });
+    mockOauth.tokenResponse = null;
+  });
+
+  it("keeps the admin's in-flight endpoint edits when the token is invalidated", async () => {
+    // Regression: invalidation used to form.resetFields the endpoint fields; with the edit Form's
+    // initialValues that silently reverted an admin-corrected token_url back to the saved (wrong)
+    // value while still looking plausible. Only credentials (the minted material) may be wiped.
+    renderOAuthEdit();
+
+    const tokenUrlInput = screen.getByPlaceholderText("https://example.com/oauth/token");
+    await act(async () => {
+      fireEvent.change(tokenUrlInput, { target: { value: "https://corrected.example.com/token" } });
+    });
+
+    act(() => {
+      mockOauth.onTokenReceived?.({ access_token: "tok-1" });
+    });
+    mockOauth.reset.mockClear();
+
+    const urlInput = screen.getByPlaceholderText("https://your-mcp-server.com");
+    await act(async () => {
+      fireEvent.change(urlInput, { target: { value: "https://moved.example.com/mcp" } });
+    });
+
+    await waitFor(() => expect(mockOauth.reset).toHaveBeenCalled());
+    expect((screen.getByPlaceholderText("https://example.com/oauth/token") as HTMLInputElement).value).toBe(
+      "https://corrected.example.com/token",
+    );
+  });
+
+  it("keeps a session-authorized token on an http to sse switch with the same url", async () => {
+    // Same url means the same resource/audience (RFC 8707): the minted token is still valid, so a
+    // pure transport swap between the two MCP wire protocols must not force a re-authorize.
+    renderOAuthEdit();
+
+    act(() => {
+      mockOauth.onTokenReceived?.({ access_token: "tok-1" });
+    });
+    mockOauth.reset.mockClear();
+
+    await selectAntOption("Transport Type", "Server-Sent Events (SSE)");
+
+    expect(mockOauth.reset).not.toHaveBeenCalled();
+    expect(mockRemoveToken).not.toHaveBeenCalled();
+  });
+});
+
 describe("MCPServerEdit (tool allowlist)", () => {
   beforeEach(() => {
     vi.clearAllMocks();
@@ -883,6 +1115,55 @@ describe("MCPServerEdit (tool list fetch)", () => {
     expect(mockGetToken).toHaveBeenCalledWith("oauth_server_1", "user-1");
   });
 
+  it("forwards the sessionStorage token as the x-mcp header for an oauth_delegate server", async () => {
+    mockIsTokenValid.mockReturnValue(true);
+    mockGetToken.mockReturnValue({ access_token: "browser-token" });
+
+    render(
+      ,
+    );
+
+    await waitFor(() => {
+      expect(networking.listMCPTools).toHaveBeenCalledWith(
+        "access-token",
+        "oauth_server_1",
+        { "x-mcp-oauth_server-authorization": "Bearer browser-token" },
+        true,
+      );
+    });
+    expect(mockGetToken).toHaveBeenCalledWith("oauth_server_1", "user-1");
+  });
+
+  it("prompts for the browser-only authorize when a true_passthrough server has no token", async () => {
+    mockIsTokenValid.mockReturnValue(false);
+
+    render(
+      ,
+    );
+
+    await waitFor(() => {
+      expect(screen.getByTestId("mcp-tool-config").getAttribute("data-external-error")).toContain(
+        "Authorize with the upstream (browser-only",
+      );
+    });
+    expect(networking.listMCPTools).not.toHaveBeenCalled();
+    expect(screen.getByRole("button", { name: "Authorize & Fetch Tools (browser-only)" })).toBeInTheDocument();
+  });
+
   it("uses the staged OAuth token to load passthrough tools after authorize", async () => {
     const passthroughServer = { ...interactiveOAuthServer, delegate_auth_to_upstream: true };
     mockIsTokenValid.mockReturnValue(false);
@@ -1055,6 +1336,77 @@ describe("MCPServerEdit (OAuth token persistence on save)", () => {
     expect(onSuccess).not.toHaveBeenCalled();
   });
 
+  it.each([["true_passthrough"], ["oauth_delegate"]])(
+    "persists the staged token to sessionStorage on save for the %s mode",
+    async (authType) => {
+      // Regression: the save path classified the staged token with getMcpOAuthMode, which returns
+      // null for the client-forwarded modes, so setToken was never called and the browser-held
+      // token was dropped on save; the create form's submit path already committed it.
+      mockOauth.tokenResponse = { access_token: "cf-tok", expires_in: 1800, token_type: "bearer" };
+      vi.mocked(networking.updateMCPServer).mockResolvedValue({
+        ...interactiveOAuthServer,
+        auth_type: authType,
+      });
+
+      render(
+        ,
+      );
+
+      await act(async () => {
+        fireEvent.click(screen.getAllByRole("button", { name: "Save Changes" })[0]);
+      });
+
+      await waitFor(() => {
+        expect(mockSetToken).toHaveBeenCalledWith(
+          "oauth_server_1",
+          expect.objectContaining({ access_token: "cf-tok" }),
+          "user-1",
+        );
+      });
+      expect(networking.storeMCPOAuthUserCredential).not.toHaveBeenCalled();
+      const [, payload] = vi.mocked(networking.updateMCPServer).mock.calls[0];
+      expect(payload.credentials).toBeUndefined();
+      expect(JSON.stringify(payload)).not.toContain("cf-tok");
+    },
+  );
+
+  it("forwards a newly authorized browser-held token for tool loading before the form is saved", async () => {
+    // Regression: fetchTools keyed the browser-held decision off the saved mcpServer.auth_type, so
+    // after switching the form to true_passthrough and authorizing, the fresh token was not sent as
+    // the x-mcp header until the server was saved.
+    vi.mocked(networking.listMCPTools).mockResolvedValue({ tools: [], error: null });
+    mockIsTokenValid.mockReturnValue(false);
+
+    render(
+      ,
+    );
+
+    await selectAntOption("Authentication", "OAuth Delegate (client-supplied upstream token)");
+    mockOauth.tokenResponse = { access_token: "fresh-tok", token_type: "bearer" };
+    await selectAntOption("Authentication", "True Passthrough (no LiteLLM auth)");
+
+    await waitFor(() => {
+      const withHeaders = vi
+        .mocked(networking.listMCPTools)
+        .mock.calls.find(([, , headers]) => headers && JSON.stringify(headers).includes("fresh-tok"));
+      expect(withHeaders).toBeTruthy();
+    });
+  });
+
   it("persists the passthrough token to sessionStorage on save after authorize", async () => {
     mockOauth.tokenResponse = { access_token: "pt-tok", expires_in: 1800, token_type: "bearer" };
     vi.mocked(networking.updateMCPServer).mockResolvedValue({
diff --git a/ui/litellm-dashboard/src/components/mcp_tools/mcp_server_edit.tsx b/ui/litellm-dashboard/src/components/mcp_tools/mcp_server_edit.tsx
index 70632b459fc..7446c96c40e 100644
--- a/ui/litellm-dashboard/src/components/mcp_tools/mcp_server_edit.tsx
+++ b/ui/litellm-dashboard/src/components/mcp_tools/mcp_server_edit.tsx
@@ -4,6 +4,10 @@ import { InfoCircleOutlined } from "@ant-design/icons";
 import { Button, TabGroup, TabList, Tab, TabPanels, TabPanel } from "@tremor/react";
 import {
   AUTH_TYPE,
+  isClientForwardedTokenMode,
+  getOAuthAuthorizationIdentity,
+  CLEARED_ON_INVALIDATION,
+  isHeldOAuthTokenStale,
   OAUTH_FLOW,
   MCP_OAUTH2_FLOW_M2M,
   MCP_OAUTH2_FLOW_INTERACTIVE,
@@ -13,11 +17,13 @@ import {
   getMcpOAuthMode,
   oauth2FlowToFormValue,
 } from "./types";
-import { updateMCPServer, listMCPTools, storeMCPOAuthUserCredential } from "../networking";
-import { getToken, isTokenValid, setToken } from "@/utils/mcpTokenStore";
+import { updateMCPServer, listMCPTools, storeMCPOAuthUserCredential, testMCPToolsListRequest } from "../networking";
+import { getToken, isTokenValid, removeToken, setToken } from "@/utils/mcpTokenStore";
 import { buildMcpPassthroughAuthHeader } from "@/utils/mcpHeaderUtils";
 import MCPServerCostConfig from "./mcp_server_cost_config";
 import MCPPermissionManagement from "./MCPPermissionManagement";
+import TruePassthroughWarning from "./TruePassthroughWarning";
+import PassthroughAuthorizeSection from "./PassthroughAuthorizeSection";
 import MCPToolConfiguration from "./mcp_tool_configuration";
 import StdioConfiguration from "./StdioConfiguration";
 import TokenExchangeFormFields from "./TokenExchangeFormFields";
@@ -128,11 +134,22 @@ const MCPServerEdit: React.FC = ({
     }
   };
 
+  // The auth mode every decision must key off: the admin's in-flight form selection wins over the
+  // saved record, so authorizing, loading tools, and saving all agree with what the form shows. Paths
+  // that read only mcpServer.auth_type go stale the moment the admin switches modes in the form.
+  const getEffectiveAuthType = () => form.getFieldValue("auth_type") ?? mcpServer.auth_type;
+
+  // The OAuth authorization identity (see getOAuthAuthorizationIdentity) captured when a token is fetched
+  // in this edit session; undefined when none is held. If a mint-relevant field later diverges from it,
+  // the held token (hook response + sessionStorage) is discarded so the admin must re-authorize.
+  const authorizedIdentityRef = React.useRef(undefined);
+
   const {
     startOAuthFlow,
     status: oauthStatus,
     error: oauthError,
     tokenResponse: oauthTokenResponse,
+    reset: resetOAuthFlow,
   } = useMcpOAuthFlow({
     accessToken,
     getCredentials: () => form.getFieldValue("credentials"),
@@ -161,7 +178,7 @@ const MCPServerEdit: React.FC = ({
         description: values.description || mcpServer.description,
         url,
         transport,
-        auth_type: AUTH_TYPE.OAUTH2,
+        auth_type: isClientForwardedTokenMode(values.auth_type) ? values.auth_type : AUTH_TYPE.OAUTH2,
         credentials: values.credentials,
         mcp_access_groups: values.mcp_access_groups || mcpServer.mcp_access_groups,
         static_headers: staticHeaders,
@@ -171,20 +188,39 @@ const MCPServerEdit: React.FC = ({
       };
     },
     onTokenReceived: (token) => {
-      if (token?.access_token) {
-        const credentials = {
-          access_token: token.access_token,
-          ...(token.refresh_token && { refresh_token: token.refresh_token }),
-          ...(token.expires_in && { expires_in: token.expires_in }),
-          ...(token.scope && { scope: token.scope }),
-        };
-
-        form.setFieldsValue({ credentials });
-
-        NotificationsManager.success(
-          "OAuth authorization successful! Please click 'Update MCP Server' to save the credentials.",
-        );
+      if (!token?.access_token) {
+        return;
       }
+
+      authorizedIdentityRef.current = getOAuthAuthorizationIdentity(form.getFieldsValue(true));
+      if (isClientForwardedTokenMode(getEffectiveAuthType())) {
+        const browserHeldToken = {
+          access_token: token.access_token,
+          expires_in: token.expires_in,
+          refresh_token: token.refresh_token,
+          token_type: token.token_type,
+        };
+        setToken(mcpServer.server_id, browserHeldToken, userID);
+        NotificationsManager.success(
+          "Token held for this browser session. Tools can now be loaded and configured; nothing was saved to LiteLLM.",
+        );
+        return;
+      }
+
+      const credentials = {
+        access_token: token.access_token,
+        ...(token.refresh_token && { refresh_token: token.refresh_token }),
+        ...(token.expires_in && { expires_in: token.expires_in }),
+        ...(token.scope && { scope: token.scope }),
+      };
+
+      form.setFieldsValue({ credentials });
+      // Re-capture after writing credentials so the token is not invalidated by its own credential write.
+      authorizedIdentityRef.current = getOAuthAuthorizationIdentity(form.getFieldsValue(true));
+
+      NotificationsManager.success(
+        "OAuth authorization successful! Please click 'Update MCP Server' to save the credentials.",
+      );
     },
     onBeforeRedirect: persistEditUiState,
     flowSource: "edit",
@@ -354,6 +390,91 @@ const MCPServerEdit: React.FC = ({
     // eslint-disable-next-line react-hooks/exhaustive-deps
   }, [mcpServer, accessToken, userID, oauthTokenResponse?.access_token]);
 
+  // Invalidate a token authorized in this edit session once any mint-relevant field diverges from the
+  // identity it was minted against (url, auth_type, oauth_flow_type, client creds/scopes, or the
+  // authorization/token/registration endpoints — see getOAuthAuthorizationIdentity). Discards the hook
+  // token (resetOAuthFlow, which re-runs fetchTools to prompt a fresh authorize), the sessionStorage
+  // token (removeToken, browser-held modes), and the fetched token/DCR client in the shared
+  // CLEARED_ON_INVALIDATION form fields; the admin's in-flight edit is re-applied so it is never wiped.
+  // Only fires when a token was actually authorized here (ref set), so a token already valid for the
+  // saved server on mount is left untouched. Driven from onValuesChange for user input, plus an explicit
+  // recheck after programmatic setFieldsValue paths (handleTransportChange), which antd does not report
+  // through onValuesChange.
+  const clearHeldOAuthToken = (changedValues: Record = {}) => {
+    authorizedIdentityRef.current = undefined;
+    if (mcpServer.server_id) {
+      removeToken(mcpServer.server_id, userID);
+    }
+    setTools([]);
+    resetOAuthFlow();
+    form.resetFields([...CLEARED_ON_INVALIDATION]);
+    const preserved = Object.fromEntries(
+      CLEARED_ON_INVALIDATION.filter((key) => key in changedValues).map((key) => [key, changedValues[key]]),
+    );
+    if (Object.keys(preserved).length > 0) {
+      form.setFieldsValue(preserved);
+    }
+  };
+
+  const handleFormValuesChange = (changedValues: Record) => {
+    if (isHeldOAuthTokenStale(form.getFieldsValue(true), authorizedIdentityRef.current)) {
+      clearHeldOAuthToken(changedValues);
+    }
+  };
+
+  // A token authorized in this edit session for interactive OAuth (authorization_code) is only
+  // committed to the DB on save, so a plain by-server_id listing cannot use it and the preview would
+  // stay empty until the admin saves; the create form previews the identical state through the
+  // config-based preview endpoint, which takes the staged token explicitly. Returns false when there
+  // is no staged interactive token so fetchTools falls through to the by-server_id listing.
+  const previewWithStagedInteractiveToken = async (
+    isPassthrough: boolean,
+    isBrowserHeldTokenMode: boolean,
+  ): Promise => {
+    const stagedToken =
+      !isPassthrough && !isBrowserHeldTokenMode && getEffectiveAuthType() === AUTH_TYPE.OAUTH2
+        ? oauthTokenResponse?.access_token
+        : undefined;
+    if (!stagedToken) {
+      return false;
+    }
+    setIsLoadingTools(true);
+    setToolsError(null);
+    try {
+      const values = form.getFieldsValue(true);
+      const rawTransport = values.transport || mcpServer.transport;
+      // oauth2_flow must be explicit: the preview endpoint infers client_credentials from the
+      // inherited client_id/client_secret/token_url (common once DCR or discovery filled them) and
+      // would strip the staged bearer to preview as M2M. spec_path keeps OpenAPI servers on the
+      // spec-based preview path, mirroring the create form's config.
+      const previewConfig = {
+        server_id: mcpServer.server_id,
+        server_name: values.server_name || mcpServer.server_name || mcpServer.alias,
+        url: values.url || mcpServer.url,
+        spec_path: values.spec_path || mcpServer.spec_path,
+        transport: rawTransport === TRANSPORT.OPENAPI ? TRANSPORT.HTTP : rawTransport,
+        auth_type: AUTH_TYPE.OAUTH2,
+        oauth2_flow: MCP_OAUTH2_FLOW_INTERACTIVE,
+        authorization_url: values.authorization_url,
+        token_url: values.token_url,
+        registration_url: values.registration_url,
+      };
+      const toolsResponse = await testMCPToolsListRequest(accessToken, previewConfig, stagedToken);
+      if (toolsResponse.tools && !toolsResponse.error) {
+        setTools(toolsResponse.tools);
+      } else {
+        setTools([]);
+        setToolsError(toolsResponse.message || "Failed to load tools");
+      }
+    } catch (error) {
+      setTools([]);
+      setToolsError(error instanceof Error ? error.message : "Failed to load tools");
+    } finally {
+      setIsLoadingTools(false);
+    }
+    return true;
+  };
+
   const fetchTools = async () => {
     if (!accessToken || !mcpServer.server_id) return;
 
@@ -368,7 +489,12 @@ const MCPServerEdit: React.FC = ({
         oauth2_flow: mcpServer.oauth2_flow,
         delegate_auth_to_upstream: mcpServer.delegate_auth_to_upstream,
       }) === "passthrough";
-    if (isPassthrough) {
+    const isBrowserHeldTokenMode = isClientForwardedTokenMode(getEffectiveAuthType());
+
+    if (await previewWithStagedInteractiveToken(isPassthrough, isBrowserHeldTokenMode)) {
+      return;
+    }
+    if (isPassthrough || isBrowserHeldTokenMode) {
       const token =
         oauthTokenResponse?.access_token ??
         (isTokenValid(mcpServer.server_id, userID)
@@ -376,7 +502,11 @@ const MCPServerEdit: React.FC = ({
           : null);
       if (!token) {
         setTools([]);
-        setToolsError("Authenticate with this server in the Tools tab to load and configure its tools.");
+        setToolsError(
+          isBrowserHeldTokenMode
+            ? "Authorize with the upstream (browser-only, in the Authentication section) to load and configure this server's tools."
+            : "Authenticate with this server in the Tools tab to load and configure its tools.",
+        );
         return;
       }
       customHeaders = buildMcpPassthroughAuthHeader(mcpServer.alias, token);
@@ -439,7 +569,7 @@ const MCPServerEdit: React.FC = ({
   const handleTransportChange = (value: string) => {
     // Clear fields that are not relevant for the selected transport.
     if (value === "stdio") {
-      form.setFieldsValue({
+      const clearedForStdio = {
         url: undefined,
         spec_path: undefined,
         auth_type: undefined,
@@ -447,15 +577,17 @@ const MCPServerEdit: React.FC = ({
         authorization_url: undefined,
         token_url: undefined,
         registration_url: undefined,
-      });
+      };
+      form.setFieldsValue(clearedForStdio);
     } else if (value === TRANSPORT.OPENAPI) {
-      form.setFieldsValue({
+      const clearedForOpenapi = {
         url: undefined,
         command: undefined,
         args: undefined,
         env_json: undefined,
         stdio_config: undefined,
-      });
+      };
+      form.setFieldsValue(clearedForOpenapi);
     } else {
       form.setFieldsValue({
         spec_path: undefined,
@@ -465,6 +597,9 @@ const MCPServerEdit: React.FC = ({
         stdio_config: undefined,
       });
     }
+    if (isHeldOAuthTokenStale(form.getFieldsValue(true), authorizedIdentityRef.current)) {
+      clearHeldOAuthToken();
+    }
   };
 
   const handleSave = async (values: Record) => {
@@ -722,8 +857,9 @@ const MCPServerEdit: React.FC = ({
       const updated = await updateMCPServer(accessToken, payload);
 
       // Persist the token staged via "Authorize & Fetch" (mirrors the create flow's
-      // commit-on-submit): OBO writes the per-user token to the DB, passthrough keeps
-      // it in sessionStorage. M2M/static auth resolve server-side and need neither.
+      // commit-on-submit): OBO writes the per-user token to the DB; legacy passthrough and the
+      // client-forwarded modes (true_passthrough / oauth_delegate) keep it in sessionStorage and
+      // never in the server row. M2M/static auth resolve server-side and need neither.
       if (oauthTokenResponse?.access_token) {
         const oauthMode = getMcpOAuthMode({
           auth_type: restValues.auth_type,
@@ -733,23 +869,21 @@ const MCPServerEdit: React.FC = ({
         try {
           if (oauthMode === "authorization_code") {
             const scope = oauthTokenResponse.scope;
-            await storeMCPOAuthUserCredential(accessToken, mcpServer.server_id, {
+            const oauthCredentialPayload = {
               access_token: oauthTokenResponse.access_token,
               refresh_token: oauthTokenResponse.refresh_token,
               expires_in: oauthTokenResponse.expires_in,
               scopes: typeof scope === "string" && scope ? scope.split(" ") : undefined,
-            });
-          } else if (oauthMode === "passthrough") {
-            setToken(
-              mcpServer.server_id,
-              {
-                access_token: oauthTokenResponse.access_token,
-                expires_in: oauthTokenResponse.expires_in,
-                refresh_token: oauthTokenResponse.refresh_token,
-                token_type: oauthTokenResponse.token_type,
-              },
-              userID,
-            );
+            };
+            await storeMCPOAuthUserCredential(accessToken, mcpServer.server_id, oauthCredentialPayload);
+          } else if (oauthMode === "passthrough" || isClientForwardedTokenMode(restValues.auth_type)) {
+            const browserHeldToken = {
+              access_token: oauthTokenResponse.access_token,
+              expires_in: oauthTokenResponse.expires_in,
+              refresh_token: oauthTokenResponse.refresh_token,
+              token_type: oauthTokenResponse.token_type,
+            };
+            setToken(mcpServer.server_id, browserHeldToken, userID);
           }
         } catch (error: unknown) {
           const message = error instanceof Error ? error.message : "";
@@ -775,7 +909,13 @@ const MCPServerEdit: React.FC = ({
       
       
         
-          

+          
              = ({
 
             {/* Authentication - for HTTP, SSE, and OpenAPI */}
             {!isStdioTransport && (
-              
-                
-              
+              <>
+                
+                  
+                
+                
+                
+              
             )}
 
             {isStdioTransport && (
diff --git a/ui/litellm-dashboard/src/components/mcp_tools/mcp_tools.test.tsx b/ui/litellm-dashboard/src/components/mcp_tools/mcp_tools.test.tsx
index 3346bc342f3..2e8fa901f6d 100644
--- a/ui/litellm-dashboard/src/components/mcp_tools/mcp_tools.test.tsx
+++ b/ui/litellm-dashboard/src/components/mcp_tools/mcp_tools.test.tsx
@@ -91,6 +91,37 @@ describe("MCPToolsViewer auth gate routing", () => {
     expect(screen.queryByText(GATE_TEXT)).not.toBeInTheDocument();
   });
 
+  it.each([["true_passthrough"], ["oauth_delegate"]])(
+    "shows the Authorize gate for a %s server without a browser token and does not list tools",
+    async (authType) => {
+      renderViewer({ auth_type: authType, oauth2_flow: null, delegate_auth_to_upstream: false });
+
+      expect(await screen.findByText(GATE_TEXT)).toBeInTheDocument();
+      expect(screen.getByRole("button", { name: "Authorize" })).toBeInTheDocument();
+      expect(vi.mocked(listMCPTools)).not.toHaveBeenCalled();
+      expect(vi.mocked(getMCPOAuthUserCredentialStatus)).not.toHaveBeenCalled();
+    },
+  );
+
+  it.each([["true_passthrough"], ["oauth_delegate"]])(
+    "forwards the session token via the x-mcp header for a %s server that has one",
+    async (authType) => {
+      vi.mocked(isTokenValid).mockReturnValue(true);
+      vi.mocked(getToken).mockReturnValue({ access_token: "upstream-tok" } as ReturnType);
+
+      renderViewer({ auth_type: authType, oauth2_flow: null, delegate_auth_to_upstream: false });
+
+      await waitFor(() =>
+        expect(vi.mocked(listMCPTools)).toHaveBeenCalledWith(
+          "litellm-key",
+          "srv-1",
+          expect.objectContaining({ "x-mcp-slack-authorization": "Bearer upstream-tok" }),
+        ),
+      );
+      expect(screen.queryByText(GATE_TEXT)).not.toBeInTheDocument();
+    },
+  );
+
   it("lists tools for an OBO server when the user has a DB credential, with no x-mcp header", async () => {
     renderViewer({ oauth2_flow: null, delegate_auth_to_upstream: false });
 
diff --git a/ui/litellm-dashboard/src/components/mcp_tools/mcp_tools.tsx b/ui/litellm-dashboard/src/components/mcp_tools/mcp_tools.tsx
index e436732ba3b..928a2e3c6bb 100644
--- a/ui/litellm-dashboard/src/components/mcp_tools/mcp_tools.tsx
+++ b/ui/litellm-dashboard/src/components/mcp_tools/mcp_tools.tsx
@@ -2,7 +2,14 @@ import React, { useCallback, useEffect, useState } from "react";
 import { useQuery, useMutation } from "@tanstack/react-query";
 import { ToolTestPanel } from "./ToolTestPanel";
 import { resolveLogoSrc } from "@/lib/assetPaths";
-import { MCPTool, MCPToolsViewerProps, MCPContent, CallMCPToolResponse, getMcpOAuthMode } from "./types";
+import {
+  isClientForwardedTokenMode,
+  MCPTool,
+  MCPToolsViewerProps,
+  MCPContent,
+  CallMCPToolResponse,
+  getMcpOAuthMode,
+} from "./types";
 import { listMCPTools, callMCPTool, getMCPOAuthUserCredentialStatus } from "../networking";
 import { isTokenValid, getToken, removeToken } from "@/utils/mcpTokenStore";
 import { sanitizeMcpAliasForHeader, buildMcpPassthroughAuthHeader } from "@/utils/mcpHeaderUtils";
@@ -42,19 +49,23 @@ const MCPToolsViewer = ({
   // service token and needs no gate.
   const oauthMode = getMcpOAuthMode({ auth_type, oauth2_flow, delegate_auth_to_upstream });
   const isPassthrough = oauthMode === "passthrough";
+  // The client-forwarded token modes gate the same way as PKCE passthrough: the
+  // browser session token (established via the browser-only Authorize in the
+  // create/edit forms, or right here) is the upstream credential.
+  const usesBrowserHeldToken = isPassthrough || isClientForwardedTokenMode(auth_type);
   const isAuthorizationCode = oauthMode === "authorization_code";
   const [oauthToken, setOauthToken] = useState(() =>
-    isPassthrough && isTokenValid(serverId, userID) ? getToken(serverId, userID)?.access_token ?? null : null,
+    usesBrowserHeldToken && isTokenValid(serverId, userID) ? getToken(serverId, userID)?.access_token ?? null : null,
   );
 
   // Re-sync token when serverId/userID changes (useState initializer only runs on mount).
   useEffect(() => {
-    if (!isPassthrough) {
+    if (!usesBrowserHeldToken) {
       setOauthToken(null);
       return;
     }
     setOauthToken(isTokenValid(serverId, userID) ? getToken(serverId, userID)?.access_token ?? null : null);
-  }, [serverId, userID, isPassthrough]);
+  }, [serverId, userID, usesBrowserHeldToken]);
 
   const {
     startOAuthFlow,
@@ -109,7 +120,7 @@ const MCPToolsViewer = ({
     // x-mcp-{alias}-{header} pattern and forwards it to the upstream MCP server.
     // When no alias is available, fall back to x-mcp-auth (legacy but still supported).
     // Passthrough only: authorization_code/token_exchange/M2M tokens are attached server-side, not from the browser.
-    if (isPassthrough && oauthToken) {
+    if (usesBrowserHeldToken && oauthToken) {
       Object.assign(customHeaders, buildMcpPassthroughAuthHeader(serverAlias, oauthToken));
     }
 
@@ -164,7 +175,8 @@ const MCPToolsViewer = ({
     // Passthrough blocks until a browser session token exists; authorization_code blocks until
     // the user has a valid DB credential (else the backend returns no tools).
     enabled:
-      !!accessToken && (isPassthrough ? oauthToken !== null : isAuthorizationCode ? hasAuthorizationCodeCred : true),
+      !!accessToken &&
+      (usesBrowserHeldToken ? oauthToken !== null : isAuthorizationCode ? hasAuthorizationCodeCred : true),
     staleTime: 30000, // Consider data fresh for 30 seconds
     retry: (failureCount, error: any) => {
       // Don't retry on 401 — token is invalid, user must re-authenticate
@@ -253,7 +265,8 @@ const MCPToolsViewer = ({
   // passthrough needs a browser token; authorization_code needs a stored DB credential or a
   // still-valid one — a 401 from the list call means the backend has none even
   // after attempting a refresh, so re-authorization is required.
-  const authGateActive = (isPassthrough && !oauthToken) || authorizationCodeNeedsAuth || authorizationCodeTokenRejected;
+  const authGateActive =
+    (usesBrowserHeldToken && !oauthToken) || authorizationCodeNeedsAuth || authorizationCodeTokenRejected;
   // Treat authorization_code credential-status loading as "tools loading" so the empty state
   // doesn't flash before we know whether the user needs to authorize.
   const toolsAreaLoading = isLoadingTools || authorizationCodeStatusLoading;
@@ -359,7 +372,7 @@ const MCPToolsViewer = ({
                 
 
                 {/* Passthrough auth gate — browser session token absent */}
-                {isPassthrough && !oauthToken && (
+                {usesBrowserHeldToken && !oauthToken && (
                   

                     
                     
Authentication required

diff --git a/ui/litellm-dashboard/src/components/mcp_tools/types.test.tsx b/ui/litellm-dashboard/src/components/mcp_tools/types.test.tsx
index 846bcfc9e0b..c6faca1fb51 100644
--- a/ui/litellm-dashboard/src/components/mcp_tools/types.test.tsx
+++ b/ui/litellm-dashboard/src/components/mcp_tools/types.test.tsx
@@ -7,9 +7,36 @@ import {
   handleTransport,
   handleAuth,
   getMcpOAuthMode,
+  getOAuthAuthorizationIdentity,
+  isHeldOAuthTokenStale,
   oauth2FlowToFormValue,
 } from "./types";
 
+describe("getOAuthAuthorizationIdentity", () => {
+  // Regression: the identity used to pick the audience from spec_path only when values.transport was
+  // OPENAPI, but the create form keeps transport in component state, so values.transport was absent and
+  // spec_path edits on OpenAPI servers never invalidated a held token.
+  it("changes when spec_path changes even when transport is absent from form values", () => {
+    const authorized = { auth_type: AUTH_TYPE.OAUTH2, spec_path: "https://a.example.com/openapi.json" };
+    const edited = { auth_type: AUTH_TYPE.OAUTH2, spec_path: "https://b.example.com/openapi.json" };
+    expect(getOAuthAuthorizationIdentity(edited)).not.toBe(getOAuthAuthorizationIdentity(authorized));
+    expect(isHeldOAuthTokenStale(edited, getOAuthAuthorizationIdentity(authorized))).toBe(true);
+  });
+
+  it("changes when url changes", () => {
+    const authorized = { auth_type: AUTH_TYPE.OAUTH2, url: "https://a.example.com/mcp" };
+    const edited = { auth_type: AUTH_TYPE.OAUTH2, url: "https://b.example.com/mcp" };
+    expect(getOAuthAuthorizationIdentity(edited)).not.toBe(getOAuthAuthorizationIdentity(authorized));
+  });
+
+  it("is stable across non-mint fields", () => {
+    const authorized = { auth_type: AUTH_TYPE.OAUTH2, url: "https://a.example.com/mcp", server_name: "one" };
+    const renamed = { auth_type: AUTH_TYPE.OAUTH2, url: "https://a.example.com/mcp", server_name: "two" };
+    expect(getOAuthAuthorizationIdentity(renamed)).toBe(getOAuthAuthorizationIdentity(authorized));
+    expect(isHeldOAuthTokenStale(renamed, getOAuthAuthorizationIdentity(authorized))).toBe(false);
+  });
+});
+
 describe("handleTransport", () => {
   it("should default to SSE when transport is null", () => {
     expect(handleTransport(null)).toBe(TRANSPORT.SSE);
diff --git a/ui/litellm-dashboard/src/components/mcp_tools/types.tsx b/ui/litellm-dashboard/src/components/mcp_tools/types.tsx
index 9469d7bd89e..3eba8b30968 100644
--- a/ui/litellm-dashboard/src/components/mcp_tools/types.tsx
+++ b/ui/litellm-dashboard/src/components/mcp_tools/types.tsx
@@ -41,13 +41,71 @@ export const AUTH_TYPE = {
   OAUTH2: "oauth2",
   OAUTH2_TOKEN_EXCHANGE: "oauth2_token_exchange",
   AWS_SIGV4: "aws_sigv4",
+  TRUE_PASSTHROUGH: "true_passthrough",
+  OAUTH_DELEGATE: "oauth_delegate",
 };
 
+// The two client-forwarded token modes: the caller supplies the upstream Authorization (forwarded
+// verbatim for true_passthrough, alongside LiteLLM admission for oauth_delegate). The dashboard holds
+// their token in sessionStorage instead of persisting it, and the browser-authorize temp payload keeps
+// their real auth_type so the backend does not treat them as needing a stored per-user token.
+export const isClientForwardedTokenMode = (authType?: string | null): boolean =>
+  authType === AUTH_TYPE.TRUE_PASSTHROUGH || authType === AUTH_TYPE.OAUTH_DELEGATE;
+
 export const OAUTH_FLOW = {
   INTERACTIVE: "interactive",
   M2M: "m2m",
 };
 
+// The fields that determine which upstream OAuth token "Authorize & Fetch" mints: the resource/audience
+// (url, or spec_path for OpenAPI servers), the OAuth mode/grant (auth_type, oauth_flow_type), the OAuth
+// client and requested scope (credentials.client_id / client_secret / scopes), and the authorization-server
+// endpoints (authorization_url / token_url / registration_url). Grounded in RFC 8707 / RFC 8693 and the MCP
+// auth spec: an access token is bound to exactly this tuple (resource/audience + scope + client + issuer), so
+// a previously authorized token is stale if and only if this identity changes and must be re-minted.
+// url and spec_path are compared independently rather than selected by transport: the create form keeps
+// transport in component state, not in form values, so a transport-conditional target would silently pin the
+// audience to a missing url and never fire for spec_path edits on OpenAPI servers. Mirrors the backend's
+// mcp_oauth_token_identity. Deliberately EXCLUDES: transport itself (http<->sse on the same url is the same
+// audience; a switch to/from OpenAPI shows up as url/spec_path changes because each form clears the field the
+// new transport does not use), delegate_auth_to_upstream (a downstream-usage toggle that is never sent to the
+// authorize request), and all metadata/RBAC/routing fields. Shared by the create and edit forms so their
+// invalidation logic cannot drift.
+export const getOAuthAuthorizationIdentity = (values: Record): string => {
+  const credentials = (values.credentials ?? {}) as Record;
+  const identity = {
+    url: typeof values.url === "string" ? values.url : null,
+    spec_path: typeof values.spec_path === "string" ? values.spec_path : null,
+    auth_type: values.auth_type ?? null,
+    oauth_flow_type: values.oauth_flow_type ?? null,
+    client_id: credentials.client_id ?? null,
+    client_secret: credentials.client_secret ?? null,
+    scopes: credentials.scopes ?? null,
+    authorization_url: values.authorization_url ?? null,
+    token_url: values.token_url ?? null,
+    registration_url: values.registration_url ?? null,
+  };
+  return JSON.stringify(identity);
+};
+
+// The form fields wiped when a held OAuth token is invalidated: only `credentials`, which holds the
+// minted material (the fetched token + DCR client). The authorization/token/registration endpoint
+// fields are deliberately NOT wiped: nothing programmatic ever writes them (upstream discovery happens
+// backend-side), so they only ever hold admin input, and resetting them would wipe it (create) or
+// silently revert it to the saved record (edit, whose Form has initialValues). Shared by the create and
+// edit forms so what gets wiped cannot drift.
+export const CLEARED_ON_INVALIDATION = ["credentials"] as const;
+
+// True when a token was authorized in this session (authorizedIdentity recorded at mint time) and the
+// form's current identity no longer matches it. Every invalidation decision in both forms goes through
+// this single check: onValuesChange for user edits, and an explicit recheck after any programmatic
+// form.setFieldsValue (antd does not fire onValuesChange for those), so a missed event path cannot let a
+// stale token survive.
+export const isHeldOAuthTokenStale = (
+  values: Record,
+  authorizedIdentity: string | undefined,
+): boolean => authorizedIdentity !== undefined && getOAuthAuthorizationIdentity(values) !== authorizedIdentity;
+
 // Backend value of `oauth2_flow` that marks a machine-to-machine server. Distinct
 // from the UI-local OAUTH_FLOW.M2M ("m2m"); this is what the API actually returns.
 export const MCP_OAUTH2_FLOW_M2M = "client_credentials";
diff --git a/ui/litellm-dashboard/src/components/model_dashboard/HealthCheckComponent.tsx b/ui/litellm-dashboard/src/components/model_dashboard/HealthCheckComponent.tsx
index a2194e23b6e..6497f2686cb 100644
--- a/ui/litellm-dashboard/src/components/model_dashboard/HealthCheckComponent.tsx
+++ b/ui/litellm-dashboard/src/components/model_dashboard/HealthCheckComponent.tsx
@@ -1,5 +1,5 @@
 import React, { useState, useEffect, useRef } from "react";
-import { Title, Text, Button, Badge } from "@tremor/react";
+import { Title, Text, Button } from "@tremor/react";
 import { Modal } from "antd";
 import { Button as AntdButton } from "antd";
 import { ModelDataTable } from "./table";
@@ -468,21 +468,6 @@ const HealthCheckComponent: React.FC = ({
     onPageChange?.(page);
   };
 
-  const getStatusBadge = (status: string) => {
-    switch (status) {
-      case "healthy":
-        return healthy;
-      case "unhealthy":
-        return unhealthy;
-      case "checking":
-        return checking;
-      case "none":
-        return none;
-      default:
-        return unknown;
-    }
-  };
-
   const showErrorModal = (modelName: string, cleanedError: string, fullError: string) => {
     setSelectedErrorDetails({
       modelName,
@@ -612,7 +597,6 @@ const HealthCheckComponent: React.FC = ({
             handleModelSelection,
             handleSelectAll,
             runIndividualHealthCheck,
-            getStatusBadge,
             getDisplayModelName,
             showErrorModal,
             showSuccessModal,
diff --git a/ui/litellm-dashboard/src/components/model_dashboard/health_check_columns.tsx b/ui/litellm-dashboard/src/components/model_dashboard/health_check_columns.tsx
index 12cc984ef91..33c97236f97 100644
--- a/ui/litellm-dashboard/src/components/model_dashboard/health_check_columns.tsx
+++ b/ui/litellm-dashboard/src/components/model_dashboard/health_check_columns.tsx
@@ -3,6 +3,7 @@ import { Tooltip, Checkbox } from "antd";
 import { Text } from "@tremor/react";
 import { InformationCircleIcon, PlayIcon, RefreshIcon } from "@heroicons/react/outline";
 import { Team } from "@/components/key_team_helpers/key_list";
+import { IdCell, StatusBadge, type StatusTone } from "@/components/shared/table_cells";
 
 interface HealthCheckData {
   model_name: string;
@@ -21,6 +22,18 @@ interface HealthCheckData {
   health_full_error?: string;
 }
 
+const HEALTH_STATUS_TONES: Record = {
+  healthy: "success",
+  unhealthy: "error",
+  checking: "info",
+  none: "neutral",
+};
+
+const healthStatusBadge = (status: string): JSX.Element => {
+  const tone = HEALTH_STATUS_TONES[status];
+  return tone ?  : ;
+};
+
 interface HealthStatus {
   status: string;
   lastCheck: string;
@@ -38,7 +51,6 @@ export const healthCheckColumns = (
   handleModelSelection: (modelId: string, checked: boolean) => void,
   handleSelectAll: (checked: boolean) => void,
   runIndividualHealthCheck: (modelId: string) => void,
-  getStatusBadge: (status: string) => JSX.Element,
   getDisplayModelName: (model: any) => string,
   showErrorModal?: (modelName: string, cleanedError: string, fullError: string) => void,
   showSuccessModal?: (modelName: string, response: any) => void,
@@ -72,14 +84,7 @@ export const healthCheckColumns = (
             onChange={(e) => handleModelSelection(modelId, e.target.checked)}
             onClick={(e) => e.stopPropagation()}
           />
-          
-            
 setSelectedModelId && setSelectedModelId(model.model_info.id)}
-            >
-              {model.model_info.id}
-            

-          
+          
         

       );
     },
@@ -175,7 +180,7 @@ export const healthCheckColumns = (
 
       return (
         

-          {getStatusBadge(healthStatus.status)}
+          {healthStatusBadge(healthStatus.status)}
           {hasSuccessResponse && showSuccessModal && (
             
               
-          
+        
 e.stopPropagation()}>
+          
         

       ) : (
         "-"
@@ -370,15 +345,10 @@ export const columns = (
     minSize: 80,
     cell: ({ row }) => {
       const model = row.original;
-      return (
-        

-          {model.model_info.db_model ? "DB Model" : "Config Model"}
-        

+      return model.model_info.db_model ? (
+        
+      ) : (
+        
       );
     },
   },
diff --git a/ui/litellm-dashboard/src/components/networking.tsx b/ui/litellm-dashboard/src/components/networking.tsx
index 403647106d4..5ec2765c621 100644
--- a/ui/litellm-dashboard/src/components/networking.tsx
+++ b/ui/litellm-dashboard/src/components/networking.tsx
@@ -1929,6 +1929,7 @@ interface UiSpendLogsParams {
   api_key?: string;
   team_id?: string;
   request_id?: string;
+  session_id?: string;
   user_id?: string;
   end_user?: string;
   status_filter?: string;
diff --git a/ui/litellm-dashboard/src/components/onboarding_link.test.tsx b/ui/litellm-dashboard/src/components/onboarding_link.test.tsx
new file mode 100644
index 00000000000..039d5e250da
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/onboarding_link.test.tsx
@@ -0,0 +1,70 @@
+import { describe, it, expect } from "vitest";
+import { buildOnboardingUrl } from "./onboarding_link";
+
+describe("buildOnboardingUrl", () => {
+  it("points the invitation link at the dedicated /ui/onboarding route", () => {
+    expect(
+      buildOnboardingUrl({
+        baseUrl: "http://localhost:4000/",
+        invitationId: "inv-123",
+        hasUserSetupSso: false,
+        resetPassword: false,
+      }),
+    ).toBe("http://localhost:4000/ui/onboarding?invitation_id=inv-123");
+  });
+
+  it("preserves a server_root_path prefix before /ui/onboarding", () => {
+    expect(
+      buildOnboardingUrl({
+        baseUrl: "https://proxy.example.com/litellm",
+        invitationId: "inv-123",
+        hasUserSetupSso: false,
+        resetPassword: false,
+      }),
+    ).toBe("https://proxy.example.com/litellm/ui/onboarding?invitation_id=inv-123");
+  });
+
+  it("appends action=reset_password for the reset-password flow", () => {
+    expect(
+      buildOnboardingUrl({
+        baseUrl: "http://localhost:4000/",
+        invitationId: "inv-123",
+        hasUserSetupSso: false,
+        resetPassword: true,
+      }),
+    ).toBe("http://localhost:4000/ui/onboarding?invitation_id=inv-123&action=reset_password");
+  });
+
+  it("sends SSO users to the dashboard root, not the onboarding form", () => {
+    expect(
+      buildOnboardingUrl({
+        baseUrl: "http://localhost:4000/",
+        invitationId: "inv-123",
+        hasUserSetupSso: true,
+        resetPassword: false,
+      }),
+    ).toBe("http://localhost:4000/ui");
+  });
+
+  it("returns an empty string when no base URL is known yet", () => {
+    expect(
+      buildOnboardingUrl({
+        baseUrl: "",
+        invitationId: "inv-123",
+        hasUserSetupSso: false,
+        resetPassword: false,
+      }),
+    ).toBe("");
+  });
+
+  it("returns an empty string rather than an invitation_id=undefined link when the id is not ready", () => {
+    expect(
+      buildOnboardingUrl({
+        baseUrl: "http://localhost:4000/",
+        invitationId: undefined,
+        hasUserSetupSso: false,
+        resetPassword: false,
+      }),
+    ).toBe("");
+  });
+});
diff --git a/ui/litellm-dashboard/src/components/onboarding_link.tsx b/ui/litellm-dashboard/src/components/onboarding_link.tsx
index 7eb337a970a..0c27287a9e4 100644
--- a/ui/litellm-dashboard/src/components/onboarding_link.tsx
+++ b/ui/litellm-dashboard/src/components/onboarding_link.tsx
@@ -25,6 +25,32 @@ interface OnboardingProps {
   modalType?: "invitation" | "resetPassword";
 }
 
+export function buildOnboardingUrl({
+  baseUrl,
+  invitationId,
+  hasUserSetupSso,
+  resetPassword,
+}: {
+  baseUrl: string;
+  invitationId: string | undefined;
+  hasUserSetupSso: boolean;
+  resetPassword: boolean;
+}): string {
+  if (!baseUrl) {
+    return "";
+  }
+  const basePath = new URL(baseUrl).pathname;
+  const uiPath = basePath && basePath !== "/" ? `${basePath}/ui` : "ui";
+  if (hasUserSetupSso) {
+    return new URL(uiPath, baseUrl).toString();
+  }
+  if (!invitationId) {
+    return "";
+  }
+  const action = resetPassword ? "&action=reset_password" : "";
+  return new URL(`${uiPath}/onboarding?invitation_id=${invitationId}${action}`, baseUrl).toString();
+}
+
 export default function OnboardingModal({
   isInvitationLinkModalVisible,
   setIsInvitationLinkModalVisible,
@@ -41,24 +67,13 @@ export default function OnboardingModal({
     setIsInvitationLinkModalVisible(false);
   };
 
-  const getInvitationUrl = () => {
-    if (!baseUrl) {
-      return "";
-    }
-    const baseUrlObj = new URL(baseUrl);
-    const basePath = baseUrlObj.pathname; // This will be "/litellm" or ""
-    const path = basePath && basePath !== "/" ? `${basePath}/ui` : "ui";
-    // Get the path from the base URL
-    if (invitationLinkData?.has_user_setup_sso) {
-      return new URL(path, baseUrl).toString();
-    }
-    let urlPath = `${path}?invitation_id=${invitationLinkData?.id}`;
-    if (modalType === "resetPassword") {
-      urlPath += "&action=reset_password";
-    }
-    const url = new URL(urlPath, baseUrl).toString();
-    return url;
-  };
+  const getInvitationUrl = () =>
+    buildOnboardingUrl({
+      baseUrl,
+      invitationId: invitationLinkData?.id,
+      hasUserSetupSso: invitationLinkData?.has_user_setup_sso ?? false,
+      resetPassword: modalType === "resetPassword",
+    });
 
   return (
      = ({
       render: (_: unknown, record: Member) => {
         const orgMember =
           record.user_id != null ? (orgData.members || []).find((m) => m.user_id === record.user_id) : undefined;
-        return ${formatNumberWithCommas(orgMember?.spend ?? 0, 4)};
+        return ;
       },
     },
     {
diff --git a/ui/litellm-dashboard/src/components/pass_through_settings.tsx b/ui/litellm-dashboard/src/components/pass_through_settings.tsx
index 0fdd9c632bf..6c72b8a62ab 100644
--- a/ui/litellm-dashboard/src/components/pass_through_settings.tsx
+++ b/ui/litellm-dashboard/src/components/pass_through_settings.tsx
@@ -7,6 +7,7 @@ import AddPassThroughEndpoint from "./add_pass_through";
 import PassThroughInfoView from "./pass_through_info";
 import { DataTable } from "./view_logs/table";
 import { ColumnDef } from "@tanstack/react-table";
+import { IdCell, StatusBadge } from "@/components/shared/table_cells";
 import { Eye, EyeOff } from "lucide-react";
 import NotificationsManager from "./molecules/notifications_manager";
 
@@ -136,16 +137,7 @@ const PassThroughSettings: React.FC = ({
     {
       header: "ID",
       accessorKey: "id",
-      cell: (info: any) => (
-        
-          
 info.row.original.id && setSelectedEndpointId(info.row.original.id)}
-          >
-            {info.row.original.id}
-          

-        
-      ),
+      cell: (info: any) => ,
     },
     {
       header: "Path",
@@ -192,7 +184,9 @@ const PassThroughSettings: React.FC = ({
         

       ),
       accessorKey: "auth",
-      cell: (info: any) => {info.getValue() ? "Yes" : "No"},
+      cell: (info: any) => (
+        
+      ),
     },
     {
       header: "Headers",
@@ -263,8 +257,6 @@ const PassThroughSettings: React.FC = ({
        
}
-        getRowCanExpand={() => false}
         isLoading={false}
         noDataMessage="No pass-through endpoints configured"
       />
diff --git a/ui/litellm-dashboard/src/components/shared/charts/area_chart.test.tsx b/ui/litellm-dashboard/src/components/shared/charts/area_chart.test.tsx
new file mode 100644
index 00000000000..cd033c5ce27
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/charts/area_chart.test.tsx
@@ -0,0 +1,36 @@
+import { render } from "@testing-library/react";
+import React from "react";
+import { describe, expect, it } from "vitest";
+import { AreaChart } from "./area_chart";
+
+const data = [
+  { date: "2026-03-01", tokens: 100, requests: 10 },
+  { date: "2026-03-02", tokens: 150, requests: 12 },
+];
+
+describe("AreaChart", () => {
+  it("renders one area per category with the mapped stroke colors", () => {
+    const { container } = render(
+      ,
+    );
+
+    const curves = Array.from(container.querySelectorAll("path.recharts-area-curve"));
+    expect(curves).toHaveLength(2);
+    const strokes = new Set(curves.map((curve) => curve.getAttribute("stroke")));
+    expect(strokes).toEqual(new Set(["var(--color-blue-500, #3b82f6)", "var(--color-cyan-500, #06b6d4)"]));
+  });
+
+  it("renders a fade-out gradient fill per category", () => {
+    const { container } = render(
+      ,
+    );
+
+    const gradients = container.querySelectorAll("defs linearGradient");
+    expect(gradients).toHaveLength(2);
+    const areas = Array.from(container.querySelectorAll("path.recharts-area-area"));
+    expect(areas).toHaveLength(2);
+    for (const area of areas) {
+      expect(area.getAttribute("fill")).toMatch(/^url\(#fill-/);
+    }
+  });
+});
diff --git a/ui/litellm-dashboard/src/components/shared/charts/area_chart.tsx b/ui/litellm-dashboard/src/components/shared/charts/area_chart.tsx
new file mode 100644
index 00000000000..794baa13cf7
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/charts/area_chart.tsx
@@ -0,0 +1,92 @@
+"use client";
+
+import * as React from "react";
+import { Area, AreaChart as RechartsAreaChart, CartesianGrid, XAxis, YAxis } from "recharts";
+import { ChartContainer, ChartLegend, ChartLegendContent, ChartTooltip, type ChartConfig } from "@/components/ui/chart";
+import { cn } from "@/lib/cva.config";
+import { ValueTooltip, type ChartTooltipComponent } from "./chart_tooltip";
+import { categoryFills, type ChartColor } from "./colors";
+
+export type AreaChartProps> = {
+  data: readonly TDatum[];
+  index: string;
+  categories: readonly string[];
+  colors?: readonly ChartColor[];
+  valueFormatter?: (value: number) => string;
+  yAxisWidth?: number;
+  showLegend?: boolean;
+  showGridLines?: boolean;
+  showTooltip?: boolean;
+  customTooltip?: ChartTooltipComponent;
+  className?: string;
+  style?: React.CSSProperties;
+};
+
+export function AreaChart>({
+  data,
+  index,
+  categories,
+  colors,
+  valueFormatter,
+  yAxisWidth = 56,
+  showLegend = true,
+  showGridLines = true,
+  showTooltip = true,
+  customTooltip,
+  className,
+  style,
+}: AreaChartProps) {
+  const gradientId = React.useId().replace(/:/g, "");
+  const fills = categoryFills(categories.length, colors);
+  const config: ChartConfig = Object.fromEntries(categories.map((category) => [category, { label: category }]));
+  const TooltipContent = customTooltip ?? ValueTooltip;
+
+  return (
+    
+      
+        
+          {categories.map((category, i) => (
+            
+              
+              
+            
+          ))}
+        
+        {showGridLines && }
+        
+        
+        {showTooltip && (
+           (
+              
+            )}
+          />
+        )}
+        {showLegend && (
+          }
+          />
+        )}
+        {categories.map((category, i) => (
+          
+        ))}
+      
+    
+  );
+}
diff --git a/ui/litellm-dashboard/src/components/shared/charts/bar_chart.test.tsx b/ui/litellm-dashboard/src/components/shared/charts/bar_chart.test.tsx
new file mode 100644
index 00000000000..d5253c86c6f
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/charts/bar_chart.test.tsx
@@ -0,0 +1,119 @@
+import { fireEvent, render, screen } from "@testing-library/react";
+import React from "react";
+import { describe, expect, it, vi } from "vitest";
+import { BarChart } from "./bar_chart";
+
+const data = [
+  { date: "2026-03-01", passed: 10, blocked: 2 },
+  { date: "2026-03-02", passed: 15, blocked: 1 },
+];
+
+describe("BarChart", () => {
+  it("renders one bar series per category with the mapped tremor colors", () => {
+    const { container } = render(
+      ,
+    );
+
+    const rectangles = Array.from(container.querySelectorAll("path.recharts-rectangle"));
+    expect(rectangles).toHaveLength(4);
+    const fills = new Set(rectangles.map((rect) => rect.getAttribute("fill")));
+    expect(fills).toEqual(new Set(["var(--color-green-500, #22c55e)", "var(--color-red-500, #ef4444)"]));
+  });
+
+  it("falls back to the tremor default color cycle when no colors are passed", () => {
+    const { container } = render();
+
+    const fills = new Set(
+      Array.from(container.querySelectorAll("path.recharts-rectangle")).map((rect) => rect.getAttribute("fill")),
+    );
+    expect(fills).toEqual(new Set(["var(--color-blue-500, #3b82f6)", "var(--color-cyan-500, #06b6d4)"]));
+  });
+
+  it("fires onValueChange with the datum and clicked category", () => {
+    const onValueChange = vi.fn();
+    const { container } = render(
+      ,
+    );
+
+    const firstRect = container.querySelector("path.recharts-rectangle");
+    expect(firstRect).not.toBeNull();
+    fireEvent.click(firstRect!);
+
+    expect(onValueChange).toHaveBeenCalledTimes(1);
+    const expectedClickItem = {
+      date: "2026-03-01",
+      passed: 10,
+      blocked: 2,
+      categoryClicked: "passed",
+    };
+    expect(onValueChange).toHaveBeenCalledWith(expectedClickItem);
+  });
+
+  it("renders category labels on the y axis in vertical layout", () => {
+    render(
+      ,
+    );
+
+    expect(screen.getAllByText("alpha").length).toBeGreaterThan(0);
+    expect(screen.getAllByText("beta").length).toBeGreaterThan(0);
+  });
+
+  it("applies valueFormatter to the value axis ticks", () => {
+    render(
+       `${v} req`}
+      />,
+    );
+
+    expect(screen.getAllByText(/ req$/).length).toBeGreaterThan(0);
+  });
+
+  it("renders a legend by default, matching tremor, and hides it when showLegend is false", () => {
+    const { container, rerender } = render(
+      ,
+    );
+    expect(screen.getByText("passed")).toBeInTheDocument();
+    expect(container.querySelector(".recharts-legend-wrapper")).not.toBeNull();
+
+    rerender();
+    expect(screen.queryByText("passed")).not.toBeInTheDocument();
+  });
+
+  it("emits no per-chart style tag; colors flow through fills, not CSS vars", () => {
+    const { container } = render(
+      ,
+    );
+    expect(container.querySelector("style")).toBeNull();
+  });
+
+  it("stacks bars into a single column per index when stack is set", () => {
+    const { container } = render(
+      ,
+    );
+
+    const xPositions = Array.from(container.querySelectorAll("path.recharts-rectangle")).map(
+      (rect) => rect.getAttribute("d")?.split(",")[0],
+    );
+    expect(new Set(xPositions).size).toBe(2);
+  });
+});
diff --git a/ui/litellm-dashboard/src/components/shared/charts/bar_chart.tsx b/ui/litellm-dashboard/src/components/shared/charts/bar_chart.tsx
new file mode 100644
index 00000000000..6ee3319dc10
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/charts/bar_chart.tsx
@@ -0,0 +1,119 @@
+"use client";
+
+import * as React from "react";
+import { Bar, BarChart as RechartsBarChart, CartesianGrid, XAxis, YAxis } from "recharts";
+import { ChartContainer, ChartLegend, ChartLegendContent, ChartTooltip, type ChartConfig } from "@/components/ui/chart";
+import { cn } from "@/lib/cva.config";
+import { ValueTooltip, type ChartTooltipComponent } from "./chart_tooltip";
+import { categoryFills, type ChartColor } from "./colors";
+
+export type BarChartProps> = {
+  data: readonly TDatum[];
+  index: string;
+  categories: readonly string[];
+  colors?: readonly ChartColor[];
+  valueFormatter?: (value: number) => string;
+  stack?: boolean;
+  layout?: "horizontal" | "vertical";
+  yAxisWidth?: number;
+  tickGap?: number;
+  showLegend?: boolean;
+  showXAxis?: boolean;
+  showGridLines?: boolean;
+  showTooltip?: boolean;
+  customTooltip?: ChartTooltipComponent;
+  onValueChange?: (item: TDatum & { categoryClicked: string }) => void;
+  className?: string;
+  style?: React.CSSProperties;
+};
+
+export function BarChart>({
+  data,
+  index,
+  categories,
+  colors,
+  valueFormatter,
+  stack = false,
+  layout = "horizontal",
+  yAxisWidth = 56,
+  tickGap = 5,
+  showLegend = true,
+  showXAxis = true,
+  showGridLines = true,
+  showTooltip = true,
+  customTooltip,
+  onValueChange,
+  className,
+  style,
+}: BarChartProps) {
+  const fills = categoryFills(categories.length, colors);
+  const config: ChartConfig = Object.fromEntries(categories.map((category) => [category, { label: category }]));
+  const vertical = layout === "vertical";
+  const TooltipContent = customTooltip ?? ValueTooltip;
+
+  return (
+    
+      
+        {showGridLines && }
+        {vertical ? (
+          
+        ) : (
+          
+        )}
+        {vertical ? (
+          
+        ) : (
+          
+        )}
+        {showTooltip && (
+           (
+              
+            )}
+          />
+        )}
+        {showLegend && (
+          }
+          />
+        )}
+        {categories.map((category, i) => (
+           {
+                    if (item.payload) onValueChange({ ...item.payload, categoryClicked: category });
+                  }
+                : undefined
+            }
+          />
+        ))}
+      
+    
+  );
+}
diff --git a/ui/litellm-dashboard/src/components/shared/charts/chart_legend.test.tsx b/ui/litellm-dashboard/src/components/shared/charts/chart_legend.test.tsx
new file mode 100644
index 00000000000..889927aca43
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/charts/chart_legend.test.tsx
@@ -0,0 +1,32 @@
+import { render, screen } from "@testing-library/react";
+import React from "react";
+import { describe, expect, it } from "vitest";
+import { CustomLegend } from "./chart_legend";
+
+describe("CustomLegend", () => {
+  it("renders title-cased category names without the metrics prefix", () => {
+    render();
+
+    expect(screen.getByText("Total Tokens")).toBeInTheDocument();
+    expect(screen.getByText("Spend")).toBeInTheDocument();
+  });
+
+  it("matches colors to categories by index with theme-var values", () => {
+    const { container } = render(
+      ,
+    );
+
+    const dots = Array.from(container.querySelectorAll('span[style*="background-color"]'));
+    expect(dots[0]?.getAttribute("style")).toContain("--color-blue-500");
+    expect(dots[1]?.getAttribute("style")).toContain("--color-green-500");
+  });
+
+  it("cycles colors when there are more categories than colors", () => {
+    const { container } = render(
+      ,
+    );
+
+    const dots = Array.from(container.querySelectorAll('span[style*="background-color"]'));
+    expect(dots[2]?.getAttribute("style")).toContain("--color-blue-500");
+  });
+});
diff --git a/ui/litellm-dashboard/src/components/shared/charts/chart_legend.tsx b/ui/litellm-dashboard/src/components/shared/charts/chart_legend.tsx
new file mode 100644
index 00000000000..da252d8bf63
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/charts/chart_legend.tsx
@@ -0,0 +1,25 @@
+"use client";
+
+import * as React from "react";
+import { formatCategoryName } from "./chart_tooltip";
+import { chartColorValue, type ChartColor } from "./colors";
+
+export const CustomLegend = ({
+  categories,
+  colors,
+}: {
+  categories: readonly string[];
+  colors: readonly ChartColor[];
+}) => (
+  

+    {categories.map((category, idx) => (
+      

+        
+        
{formatCategoryName(category)}

+      

+    ))}
+  

+);
diff --git a/ui/litellm-dashboard/src/components/shared/charts/chart_tooltip.test.tsx b/ui/litellm-dashboard/src/components/shared/charts/chart_tooltip.test.tsx
new file mode 100644
index 00000000000..7afc7532760
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/charts/chart_tooltip.test.tsx
@@ -0,0 +1,101 @@
+import { render, screen } from "@testing-library/react";
+import React from "react";
+import { describe, expect, it } from "vitest";
+import { CustomTooltip, ValueTooltip, type ChartTooltipProps } from "./chart_tooltip";
+
+const metricsPayload = (
+  dataKey: string,
+  value: number,
+  color = "#3b82f6",
+): NonNullable[number] =>
+  ({
+    dataKey,
+    value,
+    color,
+    payload: {
+      date: "2026-01-15",
+      metrics: {
+        total_tokens: 1000,
+        prompt_tokens: 600,
+        completion_tokens: 400,
+        spend: 1234.567,
+        api_requests: 10,
+      },
+    },
+  }) as NonNullable[number];
+
+describe("CustomTooltip", () => {
+  it("returns null when not active or payload is empty", () => {
+    const inactive = render(
+      ,
+    );
+    expect(inactive.container.firstChild).toBeNull();
+
+    const empty = render();
+    expect(empty.container.firstChild).toBeNull();
+  });
+
+  it("renders the label and title-cased category names without the metrics prefix", () => {
+    render();
+
+    expect(screen.getByText("2026-01-15")).toBeInTheDocument();
+    expect(screen.getByText("Total Tokens")).toBeInTheDocument();
+    expect(screen.getByText("1,000")).toBeInTheDocument();
+  });
+
+  it("formats spend values as dollars with two decimals", () => {
+    render();
+
+    expect(screen.getByText("$1,234.57")).toBeInTheDocument();
+  });
+
+  it("shows N/A for metrics missing from the row payload", () => {
+    render();
+
+    expect(screen.getByText("N/A")).toBeInTheDocument();
+  });
+
+  it("uses the series color for the indicator dot", () => {
+    const { container } = render(
+      ,
+    );
+
+    const dot = container.querySelector('span[style*="background-color"]');
+    expect(dot?.getAttribute("style")).toContain("--color-blue-500");
+  });
+});
+
+describe("ValueTooltip", () => {
+  const payload = [
+    {
+      dataKey: "passed",
+      name: "passed",
+      value: 1000,
+      color: "#22c55e",
+      payload: { date: "2026-01-15", passed: 1000 },
+    } as NonNullable[number],
+  ];
+
+  it("returns null when not active", () => {
+    const { container } = render();
+    expect(container.firstChild).toBeNull();
+  });
+
+  it("renders label, series name, and locale-formatted value by default", () => {
+    render();
+
+    expect(screen.getByText("2026-01-15")).toBeInTheDocument();
+    expect(screen.getByText("passed")).toBeInTheDocument();
+    expect(screen.getByText("1,000")).toBeInTheDocument();
+  });
+
+  it("applies the valueFormatter to values", () => {
+    render( `$${v}`} />);
+
+    expect(screen.getByText("$1000")).toBeInTheDocument();
+  });
+});
diff --git a/ui/litellm-dashboard/src/components/shared/charts/chart_tooltip.tsx b/ui/litellm-dashboard/src/components/shared/charts/chart_tooltip.tsx
new file mode 100644
index 00000000000..2644b8f720c
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/charts/chart_tooltip.tsx
@@ -0,0 +1,97 @@
+"use client";
+
+import * as React from "react";
+import type { TooltipContentProps, TooltipValueType } from "recharts";
+
+export type ChartTooltipProps = Pick<
+  TooltipContentProps,
+  "active" | "payload" | "label"
+>;
+
+export type ChartTooltipComponent = React.ComponentType;
+
+export const formatCategoryName = (name: string): string =>
+  name
+    .replace("metrics.", "")
+    .replace(/_/g, " ")
+    .split(" ")
+    .map((word) => word.charAt(0).toUpperCase() + word.slice(1))
+    .join(" ");
+
+export const ValueTooltip = ({
+  active,
+  payload,
+  label,
+  valueFormatter,
+}: ChartTooltipProps & { valueFormatter?: (value: number) => string }) => {
+  if (!active || !payload || payload.length === 0) return null;
+
+  const formatValue = (value: unknown): string => {
+    if (typeof value === "number") return valueFormatter ? valueFormatter(value) : value.toLocaleString();
+    return value == null ? "" : String(value);
+  };
+
+  return (
+    

+      {label != null && 
{String(label)}
}
+      

+        {payload.map((item, idx) => (
+          

+            

+              
+              {String(item.name ?? item.dataKey ?? "")}
+            

+            {formatValue(item.value)}
+          

+        ))}
+      

+    

+  );
+};
+
+const rawMetricValue = (row: unknown, dataKey: string): number | undefined => {
+  if (typeof row !== "object" || row === null || !("metrics" in row)) return undefined;
+  const metrics = (row as { metrics: unknown }).metrics;
+  if (typeof metrics !== "object" || metrics === null) return undefined;
+  const metricKey = dataKey.substring(dataKey.indexOf(".") + 1);
+  const value = (metrics as Record)[metricKey];
+  return typeof value === "number" ? value : undefined;
+};
+
+const formatMetricValue = (rawValue: number | undefined, isSpend: boolean): string => {
+  if (rawValue === undefined) return "N/A";
+  if (isSpend) return `$${rawValue.toLocaleString(undefined, { minimumFractionDigits: 2, maximumFractionDigits: 2 })}`;
+  return rawValue.toLocaleString();
+};
+
+export const CustomTooltip = ({ active, payload, label }: ChartTooltipProps) => {
+  if (!active || !payload || payload.length === 0) return null;
+
+  return (
+    

+      
{label == null ? "" : String(label)}

+      {payload.map((item) => {
+        const dataKey = item.dataKey?.toString();
+        if (!dataKey || !item.payload) return null;
+
+        const formattedValue = formatMetricValue(rawMetricValue(item.payload, dataKey), dataKey.includes("spend"));
+
+        return (
+          

+            

+              
+              
{formatCategoryName(dataKey)}

+            

+            
{formattedValue}

+          

+        );
+      })}
+    

+  );
+};
diff --git a/ui/litellm-dashboard/src/components/shared/charts/colors.ts b/ui/litellm-dashboard/src/components/shared/charts/colors.ts
new file mode 100644
index 00000000000..c30f58e9e4d
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/charts/colors.ts
@@ -0,0 +1,58 @@
+export const CHART_COLOR_HEX = {
+  slate: "#64748b",
+  gray: "#6b7280",
+  zinc: "#71717a",
+  neutral: "#737373",
+  stone: "#78716c",
+  red: "#ef4444",
+  orange: "#f97316",
+  amber: "#f59e0b",
+  yellow: "#eab308",
+  lime: "#84cc16",
+  green: "#22c55e",
+  emerald: "#10b981",
+  teal: "#14b8a6",
+  cyan: "#06b6d4",
+  sky: "#0ea5e9",
+  blue: "#3b82f6",
+  indigo: "#6366f1",
+  violet: "#8b5cf6",
+  purple: "#a855f7",
+  fuchsia: "#d946ef",
+  pink: "#ec4899",
+  rose: "#f43f5e",
+} as const;
+
+export type ChartColor = keyof typeof CHART_COLOR_HEX;
+
+export const DEFAULT_COLOR_CYCLE: readonly ChartColor[] = [
+  "blue",
+  "cyan",
+  "sky",
+  "indigo",
+  "violet",
+  "purple",
+  "fuchsia",
+  "slate",
+  "gray",
+  "zinc",
+  "neutral",
+  "stone",
+  "red",
+  "orange",
+  "amber",
+  "yellow",
+  "lime",
+  "green",
+  "emerald",
+  "teal",
+  "pink",
+  "rose",
+];
+
+export const chartColorValue = (color: ChartColor): string => `var(--color-${color}-500, ${CHART_COLOR_HEX[color]})`;
+
+export const categoryFills = (count: number, colors?: readonly ChartColor[]): readonly string[] => {
+  const cycle = colors && colors.length > 0 ? colors : DEFAULT_COLOR_CYCLE;
+  return Array.from({ length: count }, (_, i) => chartColorValue(cycle[i % cycle.length]));
+};
diff --git a/ui/litellm-dashboard/src/components/shared/charts/donut_chart.test.tsx b/ui/litellm-dashboard/src/components/shared/charts/donut_chart.test.tsx
new file mode 100644
index 00000000000..123c6cad0ec
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/charts/donut_chart.test.tsx
@@ -0,0 +1,38 @@
+import { render } from "@testing-library/react";
+import React from "react";
+import { describe, expect, it } from "vitest";
+import { DonutChart } from "./donut_chart";
+
+const data = [
+  { provider: "openai", spend: 40 },
+  { provider: "anthropic", spend: 30 },
+  { provider: "bedrock", spend: 20 },
+];
+
+describe("DonutChart", () => {
+  it("renders one sector per datum, cycling the given colors", () => {
+    const { container } = render(
+      ,
+    );
+
+    const sectors = Array.from(container.querySelectorAll(".recharts-pie-sector path"));
+    expect(sectors).toHaveLength(3);
+    expect(sectors.map((sector) => sector.getAttribute("fill"))).toEqual([
+      "var(--color-cyan-500, #06b6d4)",
+      "var(--color-blue-500, #3b82f6)",
+      "var(--color-cyan-500, #06b6d4)",
+    ]);
+  });
+
+  it("renders a full pie when variant is pie and a hollow donut otherwise", () => {
+    const { container: donut } = render();
+    const { container: pie } = render(
+      ,
+    );
+
+    const donutPath = donut.querySelector(".recharts-pie-sector path")?.getAttribute("d") ?? "";
+    const piePath = pie.querySelector(".recharts-pie-sector path")?.getAttribute("d") ?? "";
+    expect(donutPath).not.toEqual(piePath);
+    expect((donutPath.match(/A/g) ?? []).length).toBeGreaterThan((piePath.match(/A/g) ?? []).length);
+  });
+});
diff --git a/ui/litellm-dashboard/src/components/shared/charts/donut_chart.tsx b/ui/litellm-dashboard/src/components/shared/charts/donut_chart.tsx
new file mode 100644
index 00000000000..c2ce8c02e35
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/charts/donut_chart.tsx
@@ -0,0 +1,67 @@
+"use client";
+
+import * as React from "react";
+import { Cell, Pie, PieChart } from "recharts";
+import { ChartContainer, ChartTooltip, type ChartConfig } from "@/components/ui/chart";
+import { cn } from "@/lib/cva.config";
+import { ValueTooltip } from "./chart_tooltip";
+import { categoryFills, type ChartColor } from "./colors";
+
+export type DonutChartProps> = {
+  data: readonly TDatum[];
+  index: string;
+  category: string;
+  colors?: readonly ChartColor[];
+  variant?: "donut" | "pie";
+  valueFormatter?: (value: number) => string;
+  showTooltip?: boolean;
+  className?: string;
+  style?: React.CSSProperties;
+};
+
+export function DonutChart>({
+  data,
+  index,
+  category,
+  colors,
+  variant = "donut",
+  valueFormatter,
+  showTooltip = true,
+  className,
+  style,
+}: DonutChartProps) {
+  const fills = categoryFills(data.length, colors);
+  const config: ChartConfig = Object.fromEntries(
+    data.map((datum, i) => {
+      const name = String(datum[index] ?? i);
+      return [name, { label: name }];
+    }),
+  );
+
+  return (
+    
+      
+        {showTooltip && (
+           (
+              
+            )}
+          />
+        )}
+        
+          {data.map((datum, i) => (
+            
+          ))}
+        
+      
+    
+  );
+}
diff --git a/ui/litellm-dashboard/src/components/shared/charts/index.ts b/ui/litellm-dashboard/src/components/shared/charts/index.ts
new file mode 100644
index 00000000000..ba0a7544ddb
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/charts/index.ts
@@ -0,0 +1,12 @@
+export { AreaChart, type AreaChartProps } from "./area_chart";
+export { BarChart, type BarChartProps } from "./bar_chart";
+export { CustomLegend } from "./chart_legend";
+export {
+  CustomTooltip,
+  ValueTooltip,
+  formatCategoryName,
+  type ChartTooltipComponent,
+  type ChartTooltipProps,
+} from "./chart_tooltip";
+export { CHART_COLOR_HEX, DEFAULT_COLOR_CYCLE, categoryFills, chartColorValue, type ChartColor } from "./colors";
+export { DonutChart, type DonutChartProps } from "./donut_chart";
diff --git a/ui/litellm-dashboard/src/components/shared/table_cells/cell_tooltip.tsx b/ui/litellm-dashboard/src/components/shared/table_cells/cell_tooltip.tsx
new file mode 100644
index 00000000000..c6e10590e8f
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/table_cells/cell_tooltip.tsx
@@ -0,0 +1,21 @@
+"use client";
+
+import * as React from "react";
+
+import { Tooltip, TooltipContent, TooltipProvider, TooltipTrigger } from "@/components/ui/tooltip";
+
+interface CellTooltipProps {
+  content: React.ReactNode;
+  trigger: React.ReactElement;
+}
+
+export function CellTooltip({ content, trigger }: CellTooltipProps) {
+  return (
+    
+      
+        
+        {content}
+      
+    
+  );
+}
diff --git a/ui/litellm-dashboard/src/components/shared/table_cells/date_cell.test.tsx b/ui/litellm-dashboard/src/components/shared/table_cells/date_cell.test.tsx
new file mode 100644
index 00000000000..719047e0cfe
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/table_cells/date_cell.test.tsx
@@ -0,0 +1,57 @@
+import { render, screen } from "@testing-library/react";
+import { describe, expect, it } from "vitest";
+
+import { DateCell, formatCellDate, formatFullTimestamp } from "./date_cell";
+
+const localIso = new Date(2026, 6, 7, 9, 50, 13).toISOString();
+
+describe("formatCellDate", () => {
+  it("formats datetime precision as 'MMM D, HH:mm:ss' without a year", () => {
+    expect(formatCellDate(new Date(2026, 6, 7, 9, 50, 13), "datetime")).toBe("Jul 7, 09:50:13");
+  });
+
+  it("zero-pads hours, minutes and seconds", () => {
+    expect(formatCellDate(new Date(2026, 0, 2, 1, 2, 3), "datetime")).toBe("Jan 2, 01:02:03");
+  });
+
+  it("formats date precision as 'MMM D, YYYY' with no time", () => {
+    expect(formatCellDate(new Date(2026, 11, 31, 23, 59, 59), "date")).toBe("Dec 31, 2026");
+  });
+});
+
+describe("formatFullTimestamp", () => {
+  it("includes year, 24h time and the IANA timezone", () => {
+    const timeZone = Intl.DateTimeFormat().resolvedOptions().timeZone;
+    expect(formatFullTimestamp(new Date(2026, 6, 7, 9, 50, 13))).toBe(`Jul 7, 2026, 09:50:13 (${timeZone})`);
+  });
+});
+
+describe("DateCell", () => {
+  it("renders the datetime format by default", () => {
+    render();
+    expect(screen.getByText("Jul 7, 09:50:13")).toBeInTheDocument();
+  });
+
+  it("renders date-only when precision is 'date'", () => {
+    render();
+    expect(screen.getByText("Jul 7, 2026")).toBeInTheDocument();
+  });
+
+  it("renders '-' for null and undefined", () => {
+    const { rerender } = render();
+    expect(screen.getByText("-")).toBeInTheDocument();
+    rerender();
+    expect(screen.getByText("-")).toBeInTheDocument();
+  });
+
+  it("renders the custom fallback for empty values", () => {
+    render();
+    expect(screen.getByText("Never")).toBeInTheDocument();
+  });
+
+  it("renders the fallback instead of 'Invalid Date' for unparseable input", () => {
+    render();
+    expect(screen.getByText("Unknown")).toBeInTheDocument();
+    expect(screen.queryByText(/Invalid/)).not.toBeInTheDocument();
+  });
+});
diff --git a/ui/litellm-dashboard/src/components/shared/table_cells/date_cell.tsx b/ui/litellm-dashboard/src/components/shared/table_cells/date_cell.tsx
new file mode 100644
index 00000000000..ee4c01bb237
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/table_cells/date_cell.tsx
@@ -0,0 +1,41 @@
+"use client";
+
+import { CellTooltip } from "./cell_tooltip";
+
+const MONTHS = ["Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"] as const;
+
+export type DatePrecision = "datetime" | "date";
+
+interface DateCellProps {
+  value: string | null | undefined;
+  precision?: DatePrecision;
+  fallback?: string;
+}
+
+const pad = (n: number): string => String(n).padStart(2, "0");
+
+export const formatCellDate = (date: Date, precision: DatePrecision): string =>
+  precision === "date"
+    ? `${MONTHS[date.getMonth()]} ${date.getDate()}, ${date.getFullYear()}`
+    : `${MONTHS[date.getMonth()]} ${date.getDate()}, ${pad(date.getHours())}:${pad(date.getMinutes())}:${pad(date.getSeconds())}`;
+
+export const formatFullTimestamp = (date: Date): string => {
+  const timeZone = Intl.DateTimeFormat().resolvedOptions().timeZone;
+  const day = `${MONTHS[date.getMonth()]} ${date.getDate()}, ${date.getFullYear()}`;
+  const time = `${pad(date.getHours())}:${pad(date.getMinutes())}:${pad(date.getSeconds())}`;
+  return `${day}, ${time} (${timeZone})`;
+};
+
+export function DateCell({ value, precision = "datetime", fallback = "-" }: DateCellProps) {
+  const date = value ? new Date(value) : null;
+  if (!date || Number.isNaN(date.getTime())) {
+    return {fallback};
+  }
+
+  return (
+    {formatCellDate(date, precision)}}
+    />
+  );
+}
diff --git a/ui/litellm-dashboard/src/components/shared/table_cells/id_cell.test.tsx b/ui/litellm-dashboard/src/components/shared/table_cells/id_cell.test.tsx
new file mode 100644
index 00000000000..1a87f17d50b
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/table_cells/id_cell.test.tsx
@@ -0,0 +1,78 @@
+import { render, screen } from "@testing-library/react";
+import userEvent from "@testing-library/user-event";
+import { describe, expect, it, vi } from "vitest";
+
+import { IdCell } from "./id_cell";
+
+const { copyToClipboardMock } = vi.hoisted(() => ({ copyToClipboardMock: vi.fn() }));
+
+vi.mock("@/utils/dataUtils", async (importOriginal) => ({
+  ...(await importOriginal()),
+  copyToClipboard: copyToClipboardMock,
+}));
+
+describe("IdCell", () => {
+  it("renders '-' for empty values", () => {
+    render();
+    expect(screen.getByText("-")).toBeInTheDocument();
+  });
+
+  it("renders the custom fallback for empty values", () => {
+    render();
+    expect(screen.getByText("—")).toBeInTheDocument();
+  });
+
+  it("renders the full id as a non-interactive pill by default", () => {
+    render();
+    const el = screen.getByText("sk-1234567890abcdef");
+    expect(el.tagName).toBe("SPAN");
+    expect(el.className).toContain("bg-blue-50");
+    expect(el.className).toContain("font-mono");
+    expect(el.className).toContain("max-w-[15ch]");
+    expect(el.className).toContain("truncate");
+  });
+
+  it("renders plain mono text without pill styling for the plain variant", () => {
+    render();
+    const el = screen.getByText("req-123");
+    expect(el.className).toContain("font-mono");
+    expect(el.className).not.toContain("bg-blue-50");
+  });
+
+  it("does not truncate when truncate is false", () => {
+    render();
+    expect(screen.getByText("audit-object-id").className).not.toContain("truncate");
+  });
+
+  it("becomes a button that fires onClick with the id value", async () => {
+    const user = userEvent.setup();
+    const onClick = vi.fn();
+    render();
+    await user.click(screen.getByRole("button", { name: "team-42" }));
+    expect(onClick).toHaveBeenCalledWith("team-42");
+  });
+
+  it("stays non-interactive when disabled, even with onClick", () => {
+    const onClick = vi.fn();
+    render();
+    expect(screen.queryByRole("button", { name: "tag-1" })).not.toBeInTheDocument();
+  });
+
+  it("copies the id via the trailing copy button without triggering row clicks", async () => {
+    const user = userEvent.setup();
+    const rowClick = vi.fn();
+    render(
+      

+        
+      
,
+    );
+    await user.click(screen.getByRole("button", { name: "Copy ID" }));
+    expect(copyToClipboardMock).toHaveBeenCalledWith("key-hash-9");
+    expect(rowClick).not.toHaveBeenCalled();
+  });
+
+  it("passes dataTestId through to the id element", () => {
+    render();
+    expect(screen.getByTestId("key-id-cell")).toHaveTextContent("k-1");
+  });
+});
diff --git a/ui/litellm-dashboard/src/components/shared/table_cells/id_cell.tsx b/ui/litellm-dashboard/src/components/shared/table_cells/id_cell.tsx
new file mode 100644
index 00000000000..6fbd2e2f9ed
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/table_cells/id_cell.tsx
@@ -0,0 +1,94 @@
+"use client";
+
+import { Copy } from "lucide-react";
+import * as React from "react";
+
+import { cn } from "@/lib/cva.config";
+import { copyToClipboard } from "@/utils/dataUtils";
+
+import { CellTooltip } from "./cell_tooltip";
+
+export type IdCellVariant = "pill" | "plain";
+
+interface IdCellProps {
+  value: string | null | undefined;
+  variant?: IdCellVariant;
+  onClick?: (value: string) => void;
+  copyable?: boolean;
+  truncate?: boolean;
+  fallback?: string;
+  tooltip?: React.ReactNode;
+  disabled?: boolean;
+  dataTestId?: string;
+  className?: string;
+}
+
+const VARIANT_CLASS: Record = {
+  pill: {
+    base: "font-mono text-xs font-normal px-2 py-0.5 rounded-md text-left bg-blue-50 text-blue-500",
+    clickable: "hover:bg-blue-100 cursor-pointer",
+  },
+  plain: {
+    base: "font-mono text-xs text-left",
+    clickable: "hover:text-blue-600 cursor-pointer",
+  },
+};
+
+export function IdCell({
+  value,
+  variant = "pill",
+  onClick,
+  copyable = false,
+  truncate = true,
+  fallback = "-",
+  tooltip,
+  disabled = false,
+  dataTestId,
+  className,
+}: IdCellProps) {
+  if (!value) {
+    return {fallback};
+  }
+
+  const clickable = !!onClick && !disabled;
+  const classes = cn(
+    VARIANT_CLASS[variant].base,
+    clickable && VARIANT_CLASS[variant].clickable,
+    truncate && "block max-w-[15ch] truncate",
+    disabled && "opacity-50",
+    className,
+  );
+
+  const idElement = clickable ? (
+    
+  ) : (
+    
+      {value}
+    
+  );
+
+  const withTooltip = ;
+
+  if (!copyable) {
+    return withTooltip;
+  }
+
+  return (
+    
+      {withTooltip}
+      
+    
+  );
+}
diff --git a/ui/litellm-dashboard/src/components/shared/table_cells/index.ts b/ui/litellm-dashboard/src/components/shared/table_cells/index.ts
new file mode 100644
index 00000000000..e189413d43d
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/table_cells/index.ts
@@ -0,0 +1,5 @@
+export { CellTooltip } from "./cell_tooltip";
+export { DateCell, formatCellDate, formatFullTimestamp, type DatePrecision } from "./date_cell";
+export { IdCell, type IdCellVariant } from "./id_cell";
+export { MoneyCell } from "./money_cell";
+export { StatusBadge, type StatusTone } from "./status_badge";
diff --git a/ui/litellm-dashboard/src/components/shared/table_cells/money_cell.test.tsx b/ui/litellm-dashboard/src/components/shared/table_cells/money_cell.test.tsx
new file mode 100644
index 00000000000..473785e31c4
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/table_cells/money_cell.test.tsx
@@ -0,0 +1,44 @@
+import { render, screen } from "@testing-library/react";
+import { describe, expect, it } from "vitest";
+
+import { MoneyCell } from "./money_cell";
+
+describe("MoneyCell", () => {
+  it("renders '-' for null and undefined", () => {
+    const { rerender } = render();
+    expect(screen.getByText("-")).toBeInTheDocument();
+    rerender();
+    expect(screen.getByText("-")).toBeInTheDocument();
+  });
+
+  it("renders the custom emptyText for null budgets", () => {
+    render();
+    expect(screen.getByText("Unlimited")).toBeInTheDocument();
+  });
+
+  it("renders '-' for zero by default", () => {
+    render();
+    expect(screen.getByText("-")).toBeInTheDocument();
+  });
+
+  it("renders a formatted zero when showZero is set, never the emptyText", () => {
+    render();
+    expect(screen.getByText("$0.00")).toBeInTheDocument();
+    expect(screen.queryByText("Unlimited")).not.toBeInTheDocument();
+  });
+
+  it("formats amounts with commas, a dollar sign and the given decimals", () => {
+    render();
+    expect(screen.getByText("$1,234.57")).toBeInTheDocument();
+  });
+
+  it("defaults to 4 decimals", () => {
+    render();
+    expect(screen.getByText("$42.0000")).toBeInTheDocument();
+  });
+
+  it("renders the sub-threshold form for amounts that round to zero", () => {
+    render();
+    expect(screen.getByText("< $0.000001")).toBeInTheDocument();
+  });
+});
diff --git a/ui/litellm-dashboard/src/components/shared/table_cells/money_cell.tsx b/ui/litellm-dashboard/src/components/shared/table_cells/money_cell.tsx
new file mode 100644
index 00000000000..9d3c747b20e
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/table_cells/money_cell.tsx
@@ -0,0 +1,23 @@
+"use client";
+
+import { formatNumberWithCommas, getSpendString } from "@/utils/dataUtils";
+
+interface MoneyCellProps {
+  value: number | null | undefined;
+  decimals?: number;
+  emptyText?: string;
+  showZero?: boolean;
+}
+
+export function MoneyCell({ value, decimals = 4, emptyText = "-", showZero = false }: MoneyCellProps) {
+  if (value === null || value === undefined || Number.isNaN(value)) {
+    return {emptyText};
+  }
+  if (value === 0) {
+    if (!showZero) {
+      return -;
+    }
+    return {`$${formatNumberWithCommas(0, decimals, false, true)}`};
+  }
+  return {getSpendString(value, decimals)};
+}
diff --git a/ui/litellm-dashboard/src/components/shared/table_cells/status_badge.test.tsx b/ui/litellm-dashboard/src/components/shared/table_cells/status_badge.test.tsx
new file mode 100644
index 00000000000..724a362c21f
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/table_cells/status_badge.test.tsx
@@ -0,0 +1,42 @@
+import { render, screen } from "@testing-library/react";
+import userEvent from "@testing-library/user-event";
+import { describe, expect, it } from "vitest";
+
+import { StatusBadge, type StatusTone } from "./status_badge";
+
+describe("StatusBadge", () => {
+  const toneClasses: Record = {
+    success: ["border-green-200", "bg-green-50", "text-green-600"],
+    error: ["border-red-200", "bg-red-50", "text-red-600"],
+    warning: ["border-amber-200", "bg-amber-50", "text-amber-600"],
+    neutral: ["border-gray-200", "bg-gray-50", "text-gray-600"],
+    info: ["border-blue-200", "bg-blue-50", "text-blue-600"],
+  };
+
+  (Object.entries(toneClasses) as [StatusTone, string[]][]).forEach(([tone, classes]) => {
+    it(`renders a tinted pill (${classes.join(" ")}) for the ${tone} tone`, () => {
+      render();
+      const badge = screen.getByText(tone);
+      classes.forEach((cls) => expect(badge.className).toContain(cls));
+    });
+  });
+
+  it("renders the label text inside an outline badge with no status dot", () => {
+    render();
+    const badge = screen.getByText("Active");
+    expect(badge.dataset.variant).toBe("outline");
+    expect(badge.querySelector("[aria-hidden]")).toBeNull();
+  });
+
+  it("passes dataTestId through", () => {
+    render();
+    expect(screen.getByTestId("key-status")).toHaveTextContent("Blocked");
+  });
+
+  it("opens the tooltip on hover, which requires Badge to forward its ref to the trigger", async () => {
+    const user = userEvent.setup();
+    render();
+    await user.hover(screen.getByText("Blocked"));
+    expect(await screen.findByText("This key was blocked by SCIM")).toBeInTheDocument();
+  });
+});
diff --git a/ui/litellm-dashboard/src/components/shared/table_cells/status_badge.tsx b/ui/litellm-dashboard/src/components/shared/table_cells/status_badge.tsx
new file mode 100644
index 00000000000..f10f817650a
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/shared/table_cells/status_badge.tsx
@@ -0,0 +1,38 @@
+"use client";
+
+import * as React from "react";
+
+import { Badge } from "@/components/ui/badge";
+import { cn } from "@/lib/cva.config";
+
+import { CellTooltip } from "./cell_tooltip";
+
+export type StatusTone = "success" | "error" | "warning" | "neutral" | "info";
+
+const TONE_CLASS: Record = {
+  success: "border-green-200 bg-green-50 text-green-600",
+  error: "border-red-200 bg-red-50 text-red-600",
+  warning: "border-amber-200 bg-amber-50 text-amber-600",
+  neutral: "border-gray-200 bg-gray-50 text-gray-600",
+  info: "border-blue-200 bg-blue-50 text-blue-600",
+};
+
+interface StatusBadgeProps {
+  tone: StatusTone;
+  label: string;
+  tooltip?: React.ReactNode;
+  dataTestId?: string;
+}
+
+export function StatusBadge({ tone, label, tooltip, dataTestId }: StatusBadgeProps) {
+  const badge = (
+    
+      {label}
+    
+  );
+
+  if (!tooltip) {
+    return badge;
+  }
+  return ;
+}
diff --git a/ui/litellm-dashboard/src/components/skill_hub_table_columns.tsx b/ui/litellm-dashboard/src/components/skill_hub_table_columns.tsx
index 8a8d7ae7042..8fc9adc75a2 100644
--- a/ui/litellm-dashboard/src/components/skill_hub_table_columns.tsx
+++ b/ui/litellm-dashboard/src/components/skill_hub_table_columns.tsx
@@ -3,6 +3,7 @@ import { Badge, Text } from "@tremor/react";
 import { Tooltip } from "antd";
 import { CopyOutlined, LinkOutlined } from "@ant-design/icons";
 import { Plugin } from "./claude_code_plugins/types";
+import { StatusBadge } from "@/components/shared/table_cells";
 
 export const skillHubColumns = (
   showModal: (skill: Plugin) => void,
@@ -104,9 +105,10 @@ export const skillHubColumns = (
     accessorKey: "enabled",
     enableSorting: true,
     cell: ({ row }) => (
-      
-        {row.original.enabled ? "Public" : "Draft"}
-      
+      
     ),
   },
 ];
diff --git a/ui/litellm-dashboard/src/components/team/TeamMemberTab.test.tsx b/ui/litellm-dashboard/src/components/team/TeamMemberTab.test.tsx
index ac0ae16a44f..a07c57eaa30 100644
--- a/ui/litellm-dashboard/src/components/team/TeamMemberTab.test.tsx
+++ b/ui/litellm-dashboard/src/components/team/TeamMemberTab.test.tsx
@@ -27,6 +27,8 @@ const mockSetSelectedEditMember = vi.fn();
 const mockSetIsEditMemberModalVisible = vi.fn();
 const mockSetIsAddMemberModalVisible = vi.fn();
 
+const budgetResetIso = new Date(2026, 6, 15, 12, 0, 0).toISOString();
+
 const createMockTeamData = (overrides: Partial = {}): TeamData => ({
   team_id: "team-123",
   team_info: {
@@ -78,6 +80,7 @@ const createMockTeamData = (overrides: Partial = {}): TeamData => ({
         rpm_limit: 100,
         model_max_budget: null,
         budget_duration: null,
+        budget_reset_at: budgetResetIso,
       },
     },
   ],
@@ -202,7 +205,7 @@ describe("TeamMembersComponent", () => {
       />,
     );
 
-    expect(screen.getByText("-")).toBeInTheDocument();
+    expect(screen.getAllByText("-").length).toBeGreaterThanOrEqual(1);
   });
 
   it("should display Default Proxy Admin tag for default_user_id", () => {
@@ -243,12 +246,12 @@ describe("TeamMembersComponent", () => {
       />,
     );
 
-    expect(screen.getByText(/\$100\.5/)).toBeInTheDocument();
+    expect(screen.getByText("$100.5000")).toBeInTheDocument();
     expect(screen.getByText(/100 RPM/)).toBeInTheDocument();
     expect(screen.getByText(/10000 TPM/)).toBeInTheDocument();
   });
 
-  it("should display No Limit for budget when member has no budget", () => {
+  it("should display the budget reset date for member with a budget reset", () => {
     renderWithProviders(
        {
       />,
     );
 
-    expect(screen.getByText("No Limit")).toBeInTheDocument();
+    expect(screen.getByText("Jul 15, 2026")).toBeInTheDocument();
+  });
+
+  it("should display formatted budget and Unlimited for member with no budget", () => {
+    renderWithProviders(
+      ,
+    );
+
+    expect(screen.getByText("$1,000.0000")).toBeInTheDocument();
+    expect(screen.getByText("Unlimited")).toBeInTheDocument();
   });
 
   it("should display No Limits for rate limits when member has no limits", () => {
diff --git a/ui/litellm-dashboard/src/components/team/TeamMemberTab.tsx b/ui/litellm-dashboard/src/components/team/TeamMemberTab.tsx
index e2f108dcbf5..b884490efc0 100644
--- a/ui/litellm-dashboard/src/components/team/TeamMemberTab.tsx
+++ b/ui/litellm-dashboard/src/components/team/TeamMemberTab.tsx
@@ -1,7 +1,7 @@
 import { useUISettings } from "@/app/(dashboard)/hooks/uiSettings/useUISettings";
 import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized";
 import { Member } from "@/components/networking";
-import { formatBudgetReset } from "@/utils/budgetUtils";
+import { DateCell, MoneyCell } from "@/components/shared/table_cells";
 import { formatNumberWithCommas } from "@/utils/dataUtils";
 import { isProxyAdminRole, isUserTeamAdminForSingleTeam } from "@/utils/roles";
 import { InfoCircleOutlined } from "@ant-design/icons";
@@ -58,14 +58,10 @@ export default function TeamMemberTab({
     return membership?.total_spend ?? 0;
   };
 
-  const getUserBudget = (userId: string | null): string | null => {
+  const getUserBudget = (userId: string | null): number | null => {
     if (!userId) return null;
     const membership = teamData.team_memberships.find((tm) => tm.user_id === userId);
-    const maxBudget = membership?.litellm_budget_table?.max_budget;
-    if (maxBudget === null || maxBudget === undefined) {
-      return null;
-    }
-    return formatNumber(maxBudget);
+    return membership?.litellm_budget_table?.max_budget ?? null;
   };
 
   // Helper function to get rate limits for a user
@@ -98,7 +94,7 @@ export default function TeamMemberTab({
   const getUserBudgetReset = (userId: string | null): string | null => {
     if (!userId) return null;
     const membership = teamData.team_memberships.find((tm) => tm.user_id === userId);
-    return formatBudgetReset(membership?.litellm_budget_table?.budget_reset_at);
+    return membership?.litellm_budget_table?.budget_reset_at ?? null;
   };
 
   const extraColumns: ColumnsType = [
@@ -146,7 +142,7 @@ export default function TeamMemberTab({
       ),
       key: "spend",
       render: (_: unknown, record: Member) => (
-        ${formatNumberWithCommas(getUserCurrentCycleSpend(record.user_id), 4)}
+        
       ),
     },
     {
@@ -159,31 +155,19 @@ export default function TeamMemberTab({
         
       ),
       key: "total_spend",
-      render: (_: unknown, record: Member) => (
-        ${formatNumberWithCommas(getUserTotalSpend(record.user_id), 4)}
-      ),
+      render: (_: unknown, record: Member) => ,
     },
     {
       title: "Team Member Budget (USD)",
       key: "budget",
-      render: (_: unknown, record: Member) => {
-        const budget = getUserBudget(record.user_id);
-        return (
-          {budget ? `$${formatNumberWithCommas(Number(budget), 4)}` : "No Limit"}
-        );
-      },
+      render: (_: unknown, record: Member) => (
+        
+      ),
     },
     {
       title: "Budget Reset",
       key: "budget_reset",
-      render: (_: unknown, record: Member) => {
-        const reset = getUserBudgetReset(record.user_id);
-        return reset ? (
-          {reset}
-        ) : (
-          —
-        );
-      },
+      render: (_: unknown, record: Member) => ,
     },
     {
       title: (
diff --git a/ui/litellm-dashboard/src/components/team/TeamVirtualKeysTable.tsx b/ui/litellm-dashboard/src/components/team/TeamVirtualKeysTable.tsx
index e51c124d618..b7128e642a5 100644
--- a/ui/litellm-dashboard/src/components/team/TeamVirtualKeysTable.tsx
+++ b/ui/litellm-dashboard/src/components/team/TeamVirtualKeysTable.tsx
@@ -2,7 +2,7 @@
 
 "use client";
 import { useKeys } from "@/app/(dashboard)/hooks/keys/useKeys";
-import { formatNumberWithCommas } from "@/utils/dataUtils";
+import { DateCell, IdCell, MoneyCell } from "@/components/shared/table_cells";
 import { ChevronDownIcon, ChevronRightIcon, ChevronUpIcon, SwitchVerticalIcon } from "@heroicons/react/outline";
 import {
   ColumnDef,
@@ -12,18 +12,7 @@ import {
   SortingState,
   useReactTable,
 } from "@tanstack/react-table";
-import {
-  Badge,
-  Button,
-  Icon,
-  Table,
-  TableBody,
-  TableCell,
-  TableHead,
-  TableHeaderCell,
-  TableRow,
-  Text,
-} from "@tremor/react";
+import { Badge, Icon, Table, TableBody, TableCell, TableHead, TableHeaderCell, TableRow, Text } from "@tremor/react";
 import { InfoCircleOutlined } from "@ant-design/icons";
 import { Popover, Skeleton, Tooltip, Typography } from "antd";
 import DefaultProxyAdminTag from "../common_components/DefaultProxyAdminTag";
@@ -214,23 +203,9 @@ export function TeamVirtualKeysTable({ teamId, teamAlias, organization }: TeamVi
         header: "Key ID",
         size: 100,
         enableSorting: true,
-        cell: (info) => {
-          const value = info.getValue() as string;
-          const width = info.cell.column.getSize();
-          return (
-            
-              
-            
-          );
-        },
+        cell: (info) => (
+           setSelectedKey(info.row.original)} />
+        ),
       },
       {
         id: "key_alias",
@@ -310,10 +285,7 @@ export function TeamVirtualKeysTable({ teamId, teamAlias, organization }: TeamVi
         header: "Created At",
         size: 120,
         enableSorting: true,
-        cell: (info) => {
-          const value = info.getValue();
-          return value ? new Date(value as string).toLocaleDateString() : "-";
-        },
+        cell: (info) => ,
       },
       {
         id: "created_by",
@@ -380,10 +352,7 @@ export function TeamVirtualKeysTable({ teamId, teamAlias, organization }: TeamVi
         header: "Updated At",
         size: 120,
         enableSorting: true,
-        cell: (info) => {
-          const value = info.getValue();
-          return value ? new Date(value as string).toLocaleDateString() : "Never";
-        },
+        cell: (info) => ,
       },
       {
         id: "last_active",
@@ -401,16 +370,7 @@ export function TeamVirtualKeysTable({ teamId, teamAlias, organization }: TeamVi
         ),
         size: 130,
         enableSorting: false,
-        cell: (info) => {
-          const value = info.getValue();
-          if (!value) return "Unknown";
-          const date = new Date(value as string);
-          return (
-            
-              {date.toLocaleDateString()}
-            
-          );
-        },
+        cell: (info) => ,
       },
       {
         id: "expires",
@@ -418,10 +378,7 @@ export function TeamVirtualKeysTable({ teamId, teamAlias, organization }: TeamVi
         header: "Expires",
         size: 120,
         enableSorting: false,
-        cell: (info) => {
-          const value = info.getValue();
-          return value ? new Date(value as string).toLocaleDateString() : "Never";
-        },
+        cell: (info) => ,
       },
       {
         id: "spend",
@@ -429,7 +386,7 @@ export function TeamVirtualKeysTable({ teamId, teamAlias, organization }: TeamVi
         header: "Spend (USD)",
         size: 100,
         enableSorting: true,
-        cell: (info) => formatNumberWithCommas(info.getValue() as number, 4),
+        cell: (info) => ,
       },
       {
         id: "max_budget",
@@ -437,11 +394,9 @@ export function TeamVirtualKeysTable({ teamId, teamAlias, organization }: TeamVi
         header: "Budget (USD)",
         size: 110,
         enableSorting: true,
-        cell: (info) => {
-          const maxBudget = info.getValue() as number | null;
-          if (maxBudget === null) return "Unlimited";
-          return `$${formatNumberWithCommas(maxBudget)}`;
-        },
+        cell: (info) => (
+          
+        ),
       },
       {
         id: "budget_reset_at",
@@ -449,10 +404,7 @@ export function TeamVirtualKeysTable({ teamId, teamAlias, organization }: TeamVi
         header: "Budget Reset",
         size: 130,
         enableSorting: false,
-        cell: (info) => {
-          const value = info.getValue();
-          return value ? new Date(value as string).toLocaleString() : "Never";
-        },
+        cell: (info) => ,
       },
       {
         id: "models",
diff --git a/ui/litellm-dashboard/src/components/ui/badge.tsx b/ui/litellm-dashboard/src/components/ui/badge.tsx
index 87b536cc37e..2e1ebffa109 100644
--- a/ui/litellm-dashboard/src/components/ui/badge.tsx
+++ b/ui/litellm-dashboard/src/components/ui/badge.tsx
@@ -21,14 +21,18 @@ const badgeVariants = cva({
   },
 });
 
-function Badge({
-  className,
-  variant = "default",
-  ...props
-}: React.ComponentProps<"span"> & VariantProps) {
-  return (
-    
-  );
-}
+const Badge = React.forwardRef<
+  HTMLSpanElement,
+  React.ComponentPropsWithoutRef<"span"> & VariantProps
+>(({ className, variant = "default", ...props }, ref) => (
+  
+));
+Badge.displayName = "Badge";
 
 export { Badge, badgeVariants };
diff --git a/ui/litellm-dashboard/src/components/ui/card.tsx b/ui/litellm-dashboard/src/components/ui/card.tsx
new file mode 100644
index 00000000000..3fc0aa65264
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/ui/card.tsx
@@ -0,0 +1,86 @@
+import * as React from "react";
+
+import { cn } from "@/lib/cva.config";
+
+const Card = React.forwardRef & { size?: "default" | "sm" }>(
+  ({ className, size = "default", ...props }, ref) => (
+    
img:first-child]:pt-0 data-[size=sm]:[--card-spacing:--spacing(4)] *:[img:first-child]:rounded-t-xl *:[img:last-child]:rounded-b-xl",
+        className,
+      )}
+      {...props}
+    />
+  ),
+);
+Card.displayName = "Card";
+
+const CardHeader = React.forwardRef>(
+  ({ className, ...props }, ref) => (
+    

+  ),
+);
+CardHeader.displayName = "CardHeader";
+
+const CardTitle = React.forwardRef>(
+  ({ className, ...props }, ref) => (
+    

+  ),
+);
+CardTitle.displayName = "CardTitle";
+
+const CardDescription = React.forwardRef>(
+  ({ className, ...props }, ref) => (
+    

+  ),
+);
+CardDescription.displayName = "CardDescription";
+
+const CardAction = React.forwardRef>(
+  ({ className, ...props }, ref) => (
+    

+  ),
+);
+CardAction.displayName = "CardAction";
+
+const CardContent = React.forwardRef>(
+  ({ className, ...props }, ref) => (
+    

+  ),
+);
+CardContent.displayName = "CardContent";
+
+const CardFooter = React.forwardRef>(
+  ({ className, ...props }, ref) => (
+    

+  ),
+);
+CardFooter.displayName = "CardFooter";
+
+export { Card, CardHeader, CardFooter, CardTitle, CardAction, CardDescription, CardContent };
diff --git a/ui/litellm-dashboard/src/components/ui/chart.test.tsx b/ui/litellm-dashboard/src/components/ui/chart.test.tsx
new file mode 100644
index 00000000000..8b70a6e3246
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/ui/chart.test.tsx
@@ -0,0 +1,38 @@
+import { render } from "@testing-library/react";
+import * as React from "react";
+import { describe, expect, it } from "vitest";
+import { ChartContainer } from "./chart";
+
+describe("ChartStyle hardening", () => {
+  it("sanitizes config keys and strips structural characters from color values", () => {
+    const { container } = render(
+      " },
+        }}
+      >
+        
+      ,
+    );
+
+    const style = container.querySelector("style");
+    expect(style).not.toBeNull();
+    const css = style!.innerHTML;
+
+    expect(css).toContain("--color-metrics_total_tokens: var(--color-blue-500, #3b82f6);");
+    expect(css).not.toContain("metrics.total_tokens");
+    expect(css).toContain("--color-evil_key:");
+    expect(css).not.toContain("<");
+    expect((css.match(/{/g) ?? []).length).toBe((css.match(/}/g) ?? []).length);
+  });
+
+  it("emits no style tag when no config entry has a color", () => {
+    const { container } = render(
+      
+        
+      ,
+    );
+    expect(container.querySelector("style")).toBeNull();
+  });
+});
diff --git a/ui/litellm-dashboard/src/components/ui/chart.tsx b/ui/litellm-dashboard/src/components/ui/chart.tsx
new file mode 100644
index 00000000000..14e10b9f06f
--- /dev/null
+++ b/ui/litellm-dashboard/src/components/ui/chart.tsx
@@ -0,0 +1,324 @@
+"use client";
+
+import * as React from "react";
+import * as RechartsPrimitive from "recharts";
+import type { TooltipValueType } from "recharts";
+
+import { cn } from "@/lib/cva.config";
+
+// Format: { THEME_NAME: CSS_SELECTOR }
+const THEMES = { light: "", dark: ".dark" } as const;
+
+const INITIAL_DIMENSION = { width: 320, height: 200 } as const;
+type TooltipNameType = number | string;
+
+export type ChartConfig = Record<
+  string,
+  {
+    label?: React.ReactNode;
+    icon?: React.ComponentType;
+  } & ({ color?: string; theme?: never } | { color?: never; theme: Record })
+>;
+
+type ChartContextProps = {
+  config: ChartConfig;
+};
+
+const ChartContext = React.createContext(null);
+
+function useChart() {
+  const context = React.useContext(ChartContext);
+
+  if (!context) {
+    throw new Error("useChart must be used within a ");
+  }
+
+  return context;
+}
+
+const ChartContainer = React.forwardRef<
+  HTMLDivElement,
+  React.ComponentPropsWithoutRef<"div"> & {
+    config: ChartConfig;
+    children: React.ComponentProps["children"];
+    initialDimension?: {
+      width: number;
+      height: number;
+    };
+  }
+>(({ id, className, children, config, initialDimension = INITIAL_DIMENSION, ...props }, ref) => {
+  const uniqueId = React.useId();
+  const chartId = `chart-${id ?? uniqueId.replace(/:/g, "")}`;
+
+  return (
+    
+      

+        
+        
+          {children}
+        
+      

+    
+  );
+});
+ChartContainer.displayName = "ChartContainer";
+
+const cssVarName = (key: string) => key.replace(/[^a-zA-Z0-9_-]/g, "_");
+const cssColorValue = (color: string) => color.replace(/[;{}<>]/g, "");
+
+const ChartStyle = ({ id, config }: { id: string; config: ChartConfig }) => {
+  const colorConfig = Object.entries(config).filter(([, config]) => config.theme ?? config.color);
+
+  if (!colorConfig.length) {
+    return null;
+  }
+
+  return (
+