scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-14 23:21:35 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 102

Merge branch 'litellm_internal_staging' into litellm_mcp_ema_subject_seam

Browse source
This commit is contained in:
tin-berri 2026-07-25 20:56:23 -07:00 • committed by GitHub
commit 358347349c
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1226 changed files with 18042 additions and 9587 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
.github/workflows/test-litellm-ui-lint.yml vendored
View file

@ -29,7 +29,15 @@ jobs:
id: changed
env:
BASE_SHA: ${{ github.event.pull_request.base.sha }}
HEAD_SHA: ${{ github.event.pull_request.head.sha }}
run: |
# base.sha is the base branch tip from when the PR was opened, while
# actions/checkout leaves HEAD on a merge of the PR into the *current*
# base tip. "$BASE_SHA"...HEAD therefore spans every base-branch commit
# landed since, so a PR that touches no UI file still gets linted
# against hundreds of other people's files. Diff the PR head against its
# own merge base instead, which is exactly what this PR changed.
merge_base=$(git merge-base "$BASE_SHA" "$HEAD_SHA")
: > "$RUNNER_TEMP/prettier_files.txt"
: > "$RUNNER_TEMP/eslint_files.txt"
while IFS= read -r f; do
@ -41,7 +49,7 @@ jobs:
*.json | *.css | *.scss | *.md | *.mdx | *.yml | *.yaml | *.html)
printf '%s\n' "$f" >> "$RUNNER_TEMP/prettier_files.txt" ;;
esac
done < <(git diff --name-only --diff-filter=ACMR --relative "$BASE_SHA"...HEAD -- .)
done < <(git diff --name-only --diff-filter=ACMR --relative "$merge_base" "$HEAD_SHA" -- .)
if [ -s "$RUNNER_TEMP/prettier_files.txt" ] || [ -s "$RUNNER_TEMP/eslint_files.txt" ]; then
echo "has_files=true" >> "$GITHUB_OUTPUT"
else

151
.github/workflows/test_server_root_path.yml vendored
View file

@ -1,151 +0,0 @@
name: Test Proxy SERVER_ROOT_PATH Routing
permissions:
contents: read
on:
pull_request:
branches:
- main
- litellm_internal_staging
- litellm_oss_staging
- "litellm_**"
jobs:
test-server-root-path:
runs-on: ubuntu-latest
timeout-minutes: 30
strategy:
fail-fast: false
matrix:
root_path: ["/api/v1", "/llmproxy"]
steps:
- name: Checkout repository
uses: actions/checkout@08eba0b27e820071cde6df949e0beb9ba4906955 # v4.3.0
with:
persist-credentials: false
- name: Free up disk space
run: |
sudo rm -rf /usr/local/lib/android /usr/share/dotnet /opt/ghc /usr/local/share/boost
sudo apt-get clean
df -h /
- name: Set up Docker Buildx
uses: docker/setup-buildx-action@8d2750c68a42422c14e847fe6c8ac0403b4cbd6f # v3.12.0
- name: Build Docker image
uses: docker/build-push-action@0adf9959216b96bec444f325f1e493d4aa344497 # v6.14.0
with:
context: .
file: ./docker/Dockerfile.non_root
tags: litellm-test:${{ github.sha }}
load: true
push: false
- name: Start LiteLLM container with SERVER_ROOT_PATH
run: |
docker run -d \
--name litellm-test \
-p 4000:4000 \
-e SERVER_ROOT_PATH="${{ matrix.root_path }}" \
-e LITELLM_MASTER_KEY="sk-1234" \
litellm-test:${{ github.sha }} \
--detailed_debug
- name: Wait for container to be healthy
run: |
echo "Waiting for LiteLLM to start..."
max_attempts=30
attempt=0
while [ $attempt -lt $max_attempts ]; do
if docker logs litellm-test 2>&1 | grep -q "Uvicorn running"; then
echo "LiteLLM started successfully"
break
fi
attempt=$((attempt + 1))
echo "Attempt $attempt/$max_attempts - waiting for server to start..."
sleep 2
done
if [ $attempt -eq $max_attempts ]; then
echo "Server failed to start within timeout"
docker logs litellm-test
exit 1
fi
sleep 5
- name: Show container logs
if: always()
run: docker logs litellm-test
- name: Test UI endpoint with root path
run: |
ROOT_PATH="${{ matrix.root_path }}"
echo "Testing UI at: http://localhost:4000${ROOT_PATH}/ui/"
for i in 1 2 3; do
content=$(curl -sL --max-time 5 -H "Authorization: Bearer sk-1234" "http://localhost:4000${ROOT_PATH}/ui/")
if echo "$content" | grep -q -E "(html|<!DOCTYPE|<head|<body)"; then
echo "UI page contains valid HTML content"
exit 0
fi
echo "Attempt $i/3 - no valid HTML, retrying in 5s..."
sleep 5
done
echo "UI page does not contain expected HTML content"
echo "Response: $content"
docker logs litellm-test
exit 1
- name: Setup Node for Playwright
uses: actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020 # v4.4.0
with:
node-version: "20"
- name: Install e2e deps and Chromium
working-directory: tests/e2e/ui
run: |
retry() {
local attempt=1
local max_attempts=4
until "$@"; do
if [ "$attempt" -ge "$max_attempts" ]; then
echo "Command failed after $attempt attempts: $*"
return 1
fi
echo "Attempt $attempt failed: $*. Retrying in $((attempt * 15))s..."
sleep $((attempt * 15))
attempt=$((attempt + 1))
done
}
npm config set fetch-retries 5
npm config set fetch-retry-mintimeout 20000
npm config set fetch-retry-maxtimeout 120000
retry npm ci
retry npx playwright install --with-deps chromium
- name: Run SERVER_ROOT_PATH redirect e2e
working-directory: tests/e2e/ui
env:
SERVER_ROOT_PATH: ${{ matrix.root_path }}
run: npx playwright test --config=serverRootPath.config.ts
- name: Upload Playwright artifacts on failure
if: failure()
uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v4.6.2
with:
name: playwright-trace-${{ strategy.job-index }}
path: tests/e2e/ui/test-results/
retention-days: 7
- name: Cleanup
if: always()
run: |
docker stop litellm-test || true
docker rm litellm-test || true

38
enterprise/litellm_enterprise/proxy/hooks/managed_files.py
View file

@ -316,26 +316,34 @@ class _PROXY_LiteLLMManagedFiles(CustomLogger, BaseFileEndpoints):
where_clause: Dict[str, Any] = {"file_purpose": "batch", **owner_filter}
if after:
where_clause["id"] = {"gt": after}
cursor_row = (
await self.prisma_client.db.litellm_managedobjecttable.find_first(
where={**where_clause, "unified_object_id": after}
)
)
if cursor_row is None:
raise HTTPException(
status_code=400,
detail=f"Invalid 'after' cursor: no batch found with id '{after}'.",
)
fetch_limit = limit or 20
if target_model_names:
# Oversample so post-fetch model-name filtering still has enough rows.
fetch_limit = max(fetch_limit * 3, 100)
page_size = limit or 20
cursor_args: Dict[str, Any] = (
{"cursor": {"unified_object_id": after}, "skip": 1} if after else {}
)
batches = await self.prisma_client.db.litellm_managedobjecttable.find_many(
where=where_clause,
take=fetch_limit,
order={"created_at": "desc"},
take=page_size + 1,
order=[{"created_at": "desc"}, {"unified_object_id": "desc"}],
**cursor_args,
)
batch_objects: List[LiteLLMBatch] = []
for batch in batches:
try:
# Stop once we have enough after filtering
if len(batch_objects) >= (limit or 20):
break
has_more = len(batches) > page_size
batch_objects: List[LiteLLMBatch] = []
for batch in batches[:page_size]:
try:
batch_data = (
json.loads(batch.file_object)
if isinstance(batch.file_object, str)
@ -351,9 +359,7 @@ class _PROXY_LiteLLMManagedFiles(CustomLogger, BaseFileEndpoints):
)
continue
return build_list_page(
batch_objects, has_more=len(batch_objects) == (limit or 20)
)
return build_list_page(batch_objects, has_more=has_more)
async def get_user_created_file_ids(
self, user_api_key_dict: UserAPIKeyAuth, model_object_ids: List[str]

4
enterprise/pyproject.toml
View file

@ -1,6 +1,6 @@
[project]
name = "litellm-enterprise"
version = "0.1.51"
version = "0.1.52"
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 = "0.1.51"
version = "0.1.52"
version_files = [
"pyproject.toml:^version",
"../pyproject.toml:litellm-enterprise==",

2
litellm-proxy-extras/litellm_proxy_extras/migrations/20260724000000_add_spend_log_tool_index_start_time_idx/migration.sql Normal file
View file

@ -0,0 +1,2 @@
-- CreateIndex
CREATE INDEX IF NOT EXISTS "LiteLLM_SpendLogToolIndex_start_time_idx" ON "LiteLLM_SpendLogToolIndex"("start_time");

1
litellm-proxy-extras/litellm_proxy_extras/schema.prisma
View file

@ -1094,6 +1094,7 @@ model LiteLLM_SpendLogToolIndex {
@@id([request_id, tool_name])
@@index([tool_name, start_time])
@@index([start_time])
}
// Prompt table for storing prompt configurations

4
litellm-proxy-extras/pyproject.toml
View file

@ -1,6 +1,6 @@
[project]
name = "litellm-proxy-extras"
version = "0.4.80"
version = "0.4.81"
description = "Additional files for the LiteLLM Proxy. Reduces the size of the main litellm package."
readme = "README.md"
requires-python = ">=3.9"
@ -26,7 +26,7 @@ required-version = ">=0.10.9"
module-root = ""
[tool.commitizen]
version = "0.4.80"
version = "0.4.81"
version_files = [
"pyproject.toml:^version",
"../pyproject.toml:litellm-proxy-extras==",

7
litellm/completion_extras/litellm_responses_transformation/transformation.py
View file

@ -35,6 +35,7 @@ from litellm.responses.sse_output_recovery import (
record_output_item_chunk,
record_output_text_chunk,
)
from litellm.responses.utils import normalize_responses_api_stream_options
from litellm.types.llms.openai import (
ChatCompletionAnnotation,
ChatCompletionReasoningItem,
@ -320,6 +321,10 @@ class LiteLLMResponsesTransformationHandler(CompletionTransformationBridge):
responses_api_request["tool_choice"] = ( # type: ignore[assignment]
self._normalize_tool_choice_for_responses_api(value)
)
elif key == "stream_options":
stream_options = normalize_responses_api_stream_options(value)
if stream_options is not None:
responses_api_request["stream_options"] = stream_options
elif key in ResponsesAPIOptionalRequestParams.__annotations__.keys():
responses_api_request[key] = value # type: ignore
elif key == "previous_response_id":
@ -360,8 +365,6 @@ class LiteLLMResponsesTransformationHandler(CompletionTransformationBridge):
continue
if key == "instructions" and instructions:
request_data["instructions"] = instructions
elif key == "stream_options" and isinstance(value, dict):
request_data["stream_options"] = value.get("include_obfuscation")
elif key == "user" and isinstance(value, str):
# OpenAI API requires user param to be max 64 chars - truncate if longer
if len(value) <= 64:

3
litellm/constants.py
View file

@ -1418,6 +1418,8 @@ LITELLM_EXPIRED_UI_SESSION_KEY_CLEANUP_BATCH_SIZE = int(
os.getenv("LITELLM_EXPIRED_UI_SESSION_KEY_CLEANUP_BATCH_SIZE", 1000)
)
LITELLM_PROXY_ADMIN_NAME = "default_user_id"
LITELLM_PROXY_BUDGET_NAME = "litellm-proxy-budget"
GLOBAL_PROXY_SPEND_CACHE_KEY = f"{LITELLM_PROXY_ADMIN_NAME}:spend"
########################### CLI SSO AUTHENTICATION CONSTANTS ###########################
LITELLM_CLI_SOURCE_IDENTIFIER = "litellm-cli"
@ -1455,6 +1457,7 @@ SPEND_LOG_CLEANUP_MAX_CONSECUTIVE_BATCH_FAILURES = int(os.getenv("SPEND_LOG_CLEA
SPEND_LOG_CLEANUP_BATCH_FAILURE_BACKOFF_SECONDS = float(
os.getenv("SPEND_LOG_CLEANUP_BATCH_FAILURE_BACKOFF_SECONDS", 0.5)
)
TOOL_SPEND_MAX_WINDOW_DAYS = 30
SPEND_LOG_PARTITION_INTERVAL = os.getenv("SPEND_LOG_PARTITION_INTERVAL", "day")
SPEND_LOG_PARTITION_PRECREATE_AHEAD = int(os.getenv("SPEND_LOG_PARTITION_PRECREATE_AHEAD", 7))
SPEND_LOG_QUEUE_SIZE_THRESHOLD = int(os.getenv("SPEND_LOG_QUEUE_SIZE_THRESHOLD", 100))

35
litellm/integrations/custom_guardrail.py
View file

@ -17,7 +17,11 @@ from typing import (
)
from litellm._logging import verbose_logger
from litellm.litellm_core_utils.core_helpers import redact_nested_match_and_regex_keys
from litellm.litellm_core_utils.core_helpers import (
get_metadata_variable_name_from_kwargs,
get_or_create_metadata_bucket,
redact_nested_match_and_regex_keys,
)
from litellm.caching import DualCache
from litellm.integrations.custom_logger import CustomLogger
from litellm.secret_managers.main import str_to_bool
@ -107,6 +111,8 @@ class CustomGuardrail(CustomLogger):
# If True, during_call runs async_moderation_hook instead of the unified apply_guardrail path.
use_native_during_call_hook: ClassVar[bool] = False
records_own_guardrail_information: ClassVar[bool] = False
def __init__(
self,
guardrail_name: Optional[str] = None,
@ -954,17 +960,8 @@ class CustomGuardrail(CustomLogger):
# should not happen
container[key] = [existing, slg]
if "metadata" in request_data:
if request_data["metadata"] is None:
request_data["metadata"] = {}
_append_guardrail_info(request_data["metadata"])
elif "litellm_metadata" in request_data:
_append_guardrail_info(request_data["litellm_metadata"])
else:
# Ensure guardrail info is always logged (e.g. proxy may not have set
# metadata yet). Attach to "metadata" so spend log / standard logging see it.
request_data["metadata"] = {}
_append_guardrail_info(request_data["metadata"])
_, metadata_bucket = get_or_create_metadata_bucket(request_data)
_append_guardrail_info(metadata_bucket)
_guardrail_self_recorded.set(True)
@ -1223,7 +1220,7 @@ def _sync_guardrail_info_to_logging_obj(request_data: dict, logging_obj: object)
"""
if logging_obj is None:
return
meta_src = request_data.get("metadata") or request_data.get("litellm_metadata") or {}
meta_src = request_data.get(get_metadata_variable_name_from_kwargs(request_data)) or {}
slg_info = meta_src.get("standard_logging_guardrail_information")
if not slg_info:
return
@ -1256,6 +1253,14 @@ def log_guardrail_information(func):
so it stays correct when guardrails run concurrently (asyncio copies the
context into each gathered task): counting shared entries would let one
guardrail's append hide another guardrail's missing record.
A guardrail that only records an entry when it actually runs (e.g.
``HeadroomGuardrail``, which returns the inputs untouched on an endpoint
whose payload it cannot act on) sets ``records_own_guardrail_information =
True`` so the auto-record is skipped even on the return paths where it
recorded nothing; otherwise a no-op early return would be logged as an
"allow"/"success" run even though the guardrail did nothing. The exception
branch below still records so a genuine failure is not lost.
"""
import functools
import inspect
@ -1291,7 +1296,7 @@ def log_guardrail_information(func):
self_recorded_token = _guardrail_self_recorded.set(False)
try:
response = await func(*args, **kwargs)
if _guardrail_self_recorded.get():
if self.records_own_guardrail_information or _guardrail_self_recorded.get():
return response
return self._process_response(
response=response,
@ -1333,7 +1338,7 @@ def log_guardrail_information(func):
self_recorded_token = _guardrail_self_recorded.set(False)
try:
response = func(*args, **kwargs)
if _guardrail_self_recorded.get():
if self.records_own_guardrail_information or _guardrail_self_recorded.get():
return response
return self._process_response(
response=response,

11
litellm/integrations/opentelemetry.py
View file

@ -883,8 +883,8 @@ class OpenTelemetry(OTELGenAISemconvMixin, CustomLogger):
request_data: dict,
parent_span: Optional[Any],
) -> None:
"""Emit ``guardrail`` spans from ``request_data["metadata"]
["standard_logging_guardrail_information"]``.
"""Emit ``guardrail`` spans from the request's proxy-internal metadata bucket
(``standard_logging_guardrail_information``).
Routed through ``_create_guardrail_span`` so the dedupe state in
``_otel_internal`` is honoured — if ``_handle_failure`` already
@ -892,7 +892,12 @@ class OpenTelemetry(OTELGenAISemconvMixin, CustomLogger):
"""
from opentelemetry import trace as _trace
metadata = (request_data or {}).get("metadata") or {}
from litellm.litellm_core_utils.core_helpers import (
get_metadata_variable_name_from_kwargs,
)
request_data = request_data or {}
metadata = request_data.get(get_metadata_variable_name_from_kwargs(request_data)) or {}
guardrail_information = metadata.get("standard_logging_guardrail_information")
if not guardrail_information:
return

11
litellm/integrations/otel/logger.py
View file

@ -17,6 +17,7 @@ from litellm.integrations.otel.model.baggage import promoted_baggage
from litellm.integrations.otel.model.config import OpenTelemetryV2Config
from litellm.integrations.otel.plumbing.context import (
is_recordable_span,
mcp_message_transport_span,
request_root_span,
resolve_mcp_span_context,
resolve_parent_context,
@ -641,8 +642,14 @@ class OpenTelemetryV2(CustomLogger):
endpoint, auth failure), so the failed request carries the same error keys
a failed LLM call does. v1's ``OpenTelemetry`` implemented this same hook;
v2 lost it when it stopped subclassing ``OpenTelemetry``, which is the
LIT-4179 regression for pre-call failures."""
span = request_root_span() or user_api_key_dict.parent_otel_span
LIT-4179 regression for pre-call failures.
An MCP message is handled on the session's task, where the request-root
anchor is whatever request opened the session, so prefer the transport the
gateway published for this specific message. Without that, a failed tool
call aimed its error at the ``initialize`` request's finished span and the
SDK dropped it, leaving the POST that actually failed unmarked."""
span = mcp_message_transport_span() or request_root_span() or user_api_key_dict.parent_otel_span
if span is None or not is_recordable_span(span):
return None
stamp_error(span, _span_error_from_exception(original_exception, traceback_str=traceback_str))

78
litellm/integrations/otel/plumbing/context.py
View file

@ -79,47 +79,67 @@ def reset_mcp_message_trace_carrier(token: "Token[Mapping[str, str] | None]") ->
_mcp_message_trace_carrier.reset(token)
# The transport span of the HTTP request carrying the CURRENT MCP message, as a
# plain ``SpanContext`` so it can cross a task boundary.
# The transport span of the HTTP request carrying the CURRENT MCP message.
#
# ``_request_root_span`` above cannot be used for MCP: a *stateful* streamable-HTTP
# session runs every message on the single task spawned by that session's
# ``initialize`` POST, so the ContextVar the ASGI request task writes at auth time
# is frozen at ``initialize`` there and never sees the later ``tools/call`` POSTs.
# Reading it from the message handler would parent every tool call in the session
# to the first request's (already ended) server span. The gateway instead resolves
# the current message's transport span on the request task and hands it over the
# same way it hands over per-request auth, and the handler publishes it here for
# the span emitter to pick up.
_mcp_message_transport_span_context: "ContextVar[SpanContext | None]" = ContextVar(
"litellm_otel_mcp_message_transport_span_context", default=None
# Reading it from the message handler parents every tool call in the session to the
# first request's server span and aims that call's ``error.*`` at it — a span that
# ended long ago, so the SDK drops the write and the failure reaches no request at
# all. The gateway instead resolves the current message's transport span on the
# request task and hands it over the same way it hands over per-request auth, and
# the handler publishes it here for the span emitter and the failure hook.
_mcp_message_transport_span: "ContextVar[Span | None]" = ContextVar(
"litellm_otel_mcp_message_transport_span", default=None
)
def set_mcp_message_transport_span_context(
span_context: "SpanContext | None",
) -> "Token[SpanContext | None]":
def set_mcp_message_transport_span(span: object) -> "Token[Span | None]":
"""Publish the transport span of the request carrying the current MCP message.
Also re-anchors the request root, so everything else the message emits or stamps
— the identity attributes seeded onto the server span, a guardrail span, a
proxy-level failure — lands on this request instead of on the one that opened
the session. The MCP SDK dispatches each message on its own task, so the anchor
is scoped to this message; the handler re-publishes it for the next one either
way. Only a transport still open for writes is anchored: replacing the anchor
with a request that already answered would just move the dropped writes from one
finished span to another.
Takes ``object`` because the gateway reads it back out of the ASGI scope, whose
values are untyped; anything that is not a usable span is stored as ``None``
rather than trusted.
Returns the reset token; the caller must reset it once the message is handled
so the transport never leaks to the next message on the same session task.
"""
return _mcp_message_transport_span_context.set(span_context)
transport = span if isinstance(span, Span) and is_recordable_span(span) else None
if transport is not None and transport.is_recording():
set_request_root_span(transport)
return _mcp_message_transport_span.set(transport)
def reset_mcp_message_transport_span_context(token: "Token[SpanContext | None]") -> None:
_mcp_message_transport_span_context.reset(token)
def reset_mcp_message_transport_span(token: "Token[Span | None]") -> None:
_mcp_message_transport_span.reset(token)
def request_root_span_context() -> "SpanContext | None":
"""The anchored request root span's context, safe to hand to another task.
def mcp_message_transport_span() -> "Span | None":
"""The published transport span, only while it is still open for writes.
A ``SpanContext`` is an immutable value, unlike the live ``Span``, so passing it
across the MCP session-task boundary cannot keep a finished span alive or invite
writes to it from the wrong request.
Recording — not merely valid — is the bar here because this span is the target
of ``error.*`` stamping from another task, and the publisher's validity check
cannot speak for a span that has since ended. A finished span keeps a valid
context forever, so it would otherwise be handed back for a write the SDK then
refuses. The POST carrying a ``tools/call`` stays open until the result is
written, so it is recording for the life of the call; a notification POST can
answer first, and this returns ``None`` for it rather than writing into the void.
"""
span = request_root_span()
return span.get_span_context() if span is not None else None
span = _mcp_message_transport_span.get()
if span is None or not span.is_recording():
return None
return span
def _mcp_transport_span_context() -> "SpanContext | None":
@ -127,12 +147,16 @@ def _mcp_transport_span_context() -> "SpanContext | None":
Prefers the transport the gateway published for this specific message; falls
back to the ambient request anchor for paths that emit an MCP span on the
request task itself (the REST MCP endpoints, the SDK).
request task itself (the REST MCP endpoints, the SDK). Parenting and linking
only need the immutable context, and unlike ``mcp_message_transport_span`` they
stay correct against a transport that has already finished, so this does not
require the span to still be recording.
"""
published = _mcp_message_transport_span_context.get()
if published is not None and published.is_valid:
return published
return request_root_span_context()
published = _mcp_message_transport_span.get()
if published is not None:
return published.get_span_context()
span = request_root_span()
return span.get_span_context() if span is not None else None
def set_request_baggage(values: Mapping[str, str], context: Context | None = None) -> Context:

19
litellm/litellm_core_utils/core_helpers.py
View file

@ -195,6 +195,25 @@ def get_metadata_variable_name_from_kwargs(
return "litellm_metadata" if "litellm_metadata" in kwargs else "metadata"
def get_or_create_metadata_bucket(
request_data: dict,
) -> tuple[Literal["metadata", "litellm_metadata"], dict]:
"""
Return the proxy-internal metadata bucket for this request, creating it if absent.
Batch/file routes store proxy state in ``litellm_metadata`` so the OpenAI
``metadata`` field can remain provider-safe (string values only). Every writer and
reader of proxy-internal metadata resolves the bucket through here, so a caller that
supplies its own ``metadata`` field cannot split them across two dicts.
"""
metadata_key = get_metadata_variable_name_from_kwargs(request_data)
metadata_bucket = request_data.get(metadata_key)
if not isinstance(metadata_bucket, dict):
metadata_bucket = {}
request_data[metadata_key] = metadata_bucket
return metadata_key, metadata_bucket
def get_litellm_metadata_from_kwargs(kwargs: dict):
"""
Helper to get litellm metadata from all litellm request kwargs

9
litellm/litellm_core_utils/litellm_logging.py
View file

@ -1878,7 +1878,14 @@ class Logging(LiteLLMLoggingBaseClass):
elif standard_logging_object is not None:
self.model_call_details["standard_logging_object"] = standard_logging_object
else:
self.model_call_details["response_cost"] = None
# Streaming reaches here before its cost is known, so the cost
# is seeded to None, but only when nothing has already
# established one. A stream that assembles into a response
# object recomputes the cost right after this; a pass-through
# stream cannot (its body is opaque) and carries the cost its
# upstream reported in the response headers, which an
# unconditional reset would discard.
self.model_call_details.setdefault("response_cost", None)
result = self._transform_usage_objects(result=result)

16
litellm/llms/anthropic/chat/guardrail_translation/handler.py
View file

@ -600,9 +600,15 @@ class AnthropicMessagesHandler(BaseTranslation):
guardrail_inputs["tool_calls"] = tool_calls_list
try:
prepared_request_data = self._prepare_request_data(
request_data,
model_response,
user_api_key_dict,
key="response",
)
_guardrailed_inputs = await guardrail_to_apply.apply_guardrail(
inputs=guardrail_inputs,
request_data=request_data if request_data is not None else {},
request_data=prepared_request_data,
input_type="response",
logging_obj=litellm_logging_obj,
)
@ -618,9 +624,15 @@ class AnthropicMessagesHandler(BaseTranslation):
string_so_far = self.get_streaming_string_so_far(responses_so_far)
try:
prepared_request_data = self._prepare_request_data(
request_data,
responses_so_far,
user_api_key_dict,
key="responses",
)
_guardrailed_inputs = await guardrail_to_apply.apply_guardrail(
inputs={"texts": [string_so_far]},
request_data=request_data if request_data is not None else {},
request_data=prepared_request_data,
input_type="response",
logging_obj=litellm_logging_obj,
)

39
litellm/llms/anthropic/experimental_pass_through/messages/interceptors/advisor.py
View file

@ -21,6 +21,7 @@ import litellm
import litellm.constants as _c
from litellm.litellm_core_utils.url_utils import validate_url
from litellm.llms.anthropic.common_utils import strip_advisor_blocks_from_messages
from litellm.router_utils.cooldown_handlers import mark_advisor_orchestration_failure
from litellm.types.llms.anthropic_messages.anthropic_response import (
AnthropicMessagesResponse,
)
@ -124,30 +125,36 @@ class AdvisorOrchestrationHandler(MessagesInterceptor):
iteration += 1
if iteration > max_uses:
raise AdvisorMaxIterationsError(
max_iterations_error = AdvisorMaxIterationsError(
f"Advisor orchestration loop exceeded max_uses={max_uses}. "
"Increase max_uses in the advisor tool definition or cap the request."
)
mark_advisor_orchestration_failure(max_iterations_error)
raise max_iterations_error
# --- Build advisor context ---
advisor_messages = _build_advisor_context(current_messages, executor_response, advisor_use_block)
# --- Advisor sub-call (always non-streaming, no tools) ---
advisor_response: AnthropicMessagesResponse = await _call_messages_handler(
model=advisor_model,
messages=advisor_messages,
tools=None,
stream=False,
max_tokens=max_tokens,
custom_llm_provider=None, # let litellm resolve from model name
metadata={
**metadata_base,
"advisor_sub_call": True,
"parent_request_id": parent_request_id,
},
api_key=advisor_api_key,
api_base=advisor_api_base,
)
try:
advisor_response: AnthropicMessagesResponse = await _call_messages_handler(
model=advisor_model,
messages=advisor_messages,
tools=None,
stream=False,
max_tokens=max_tokens,
custom_llm_provider=None, # let litellm resolve from model name
metadata={
**metadata_base,
"advisor_sub_call": True,
"parent_request_id": parent_request_id,
},
api_key=advisor_api_key,
api_base=advisor_api_base,
)
except Exception as advisor_sub_call_exception:
mark_advisor_orchestration_failure(advisor_sub_call_exception)
raise
advisor_text = _extract_response_text(advisor_response)

36
litellm/llms/bedrock/base_aws_llm.py
View file

@ -78,8 +78,10 @@ class BaseAWSLLM:
# Storage is in-process memory only: default ``DualCache()`` has no Redis backend unless attached
# elsewhere. Entry TTL: static access-key + secret + region use ``_get_default_ttl_for_boto3_credentials``
# (~59 minutes); ambient env (``_auth_with_env_vars`` returns ``ttl=None``) uses ``InMemoryCache``'s
# ``default_ttl`` (600 seconds / 10 minutes). AssumeRole, web identity, profiles, and explicit
# session-token tuples are not cached — see ``get_credentials`` and ``_get_or_set_cached_credentials``.
# ``default_ttl`` (600 seconds / 10 minutes); web identity STS credentials use
# ``_get_default_ttl_for_boto3_credentials`` (~59 minutes), keyed on all aws_* credential args
# plus ssl_verify. AssumeRole, profiles, and explicit session-token tuples are not cached — see
# ``get_credentials`` and ``_get_or_set_cached_credentials``.
_shared_iam_cache: ClassVar[DualCache] = DualCache()
def __init__(self) -> None:
@ -136,11 +138,12 @@ class BaseAWSLLM:
which ``InMemoryCache.set_cache`` resolves to ``default_ttl`` (600 seconds / 10 minutes by
default).
Used only for static access-key credentials and ambient credentials from
Used for static access-key credentials, ambient credentials from
``_auth_with_env_vars`` (including when skipping AssumeRole because the runtime identity
already matches ``aws_role_name``).
already matches ``aws_role_name``), and web identity STS credentials (plain
non-refreshable ``Credentials`` cached ~59 min, inside the 3600s STS session).
AssumeRole, web identity exchange, profiles, and explicit session-token tuples are not
AssumeRole, profiles, and explicit session-token tuples are not
cached here — shared ``Credentials`` / refresh state must not span logical sessions.
"""
cache_key = self.get_cache_key(credential_args)
@ -266,23 +269,26 @@ class BaseAWSLLM:
# Credentials - boto3.Credentials
# cache ttl - Optional[int]. If None, the credentials are not cached. Some auth flows have no expiry time.
#
# iam_cache: static keys and ambient env only (including skip-AssumeRole path).
# Do not cache AssumeRole / web identity / profile / explicit session-token paths here.
# iam_cache: static keys, ambient env (including skip-AssumeRole path), and web identity.
# Do not cache AssumeRole / profile / explicit session-token paths here.
#########################################################
if self._is_auth_with_web_identity_token(
aws_web_identity_token,
aws_role_name,
aws_session_name,
):
credentials, _cache_ttl = self._auth_with_web_identity_token(
aws_web_identity_token=cast(str, aws_web_identity_token),
aws_role_name=cast(str, aws_role_name),
aws_session_name=cast(str, aws_session_name),
aws_region_name=aws_region_name,
aws_sts_endpoint=aws_sts_endpoint,
aws_external_id=aws_external_id,
return self._get_or_set_cached_credentials(
args,
lambda: self._auth_with_web_identity_token(
aws_web_identity_token=cast(str, aws_web_identity_token),
aws_role_name=cast(str, aws_role_name),
aws_session_name=cast(str, aws_session_name),
aws_region_name=aws_region_name,
aws_sts_endpoint=aws_sts_endpoint,
aws_external_id=aws_external_id,
ssl_verify=ssl_verify,
),
)
return credentials
elif self._is_auth_with_aws_role(aws_role_name):
# Same role (IRSA/ECS/EC2): ambient creds via _get_or_set_cached_credentials like the
# default env branch; never pre-read cache (must run _is_already_running_as_role first).

23
litellm/llms/bedrock/messages/mantle_transformation.py
View file

@ -17,6 +17,10 @@ from litellm.llms.bedrock.common_utils import build_mantle_messages_url
from litellm.llms.bedrock.messages.invoke_transformations.anthropic_claude3_transformation import (
AmazonAnthropicClaudeMessagesConfig,
)
from litellm.types.llms.anthropic_messages.anthropic_response import (
AnthropicMessagesResponse,
AnthropicUsage,
)
from litellm.types.router import GenericLiteLLMParams
if TYPE_CHECKING:
@ -103,6 +107,25 @@ class AmazonMantleMessagesConfig(AmazonAnthropicClaudeMessagesConfig):
)
return {**request, "model": model_id, **stream_fields}
def transform_anthropic_messages_response(
self,
model: str,
raw_response: httpx.Response,
logging_obj: LiteLLMLoggingObj,
) -> AnthropicMessagesResponse:
response = super().transform_anthropic_messages_response(
model=model,
raw_response=raw_response,
logging_obj=logging_obj,
)
existing_usage: AnthropicUsage = response.get("usage") or AnthropicUsage()
normalized_usage: AnthropicUsage = {
"input_tokens": 0,
"output_tokens": 0,
**existing_usage,
}
return {**response, "usage": normalized_usage}
def get_async_streaming_response_iterator(
self,
model: str,

10
litellm/model_prices_and_context_window_backup.json
View file

@ -2887,7 +2887,7 @@
"input_cost_per_token": 5e-06,
"output_cost_per_token": 2.5e-05,
"litellm_provider": "azure_ai",
"max_input_tokens": 200000,
"max_input_tokens": 1000000,
"max_output_tokens": 128000,
"max_tokens": 128000,
"mode": "chat",
@ -2916,7 +2916,7 @@
"input_cost_per_token": 5e-06,
"output_cost_per_token": 2.5e-05,
"litellm_provider": "azure_ai",
"max_input_tokens": 200000,
"max_input_tokens": 1000000,
"max_output_tokens": 128000,
"max_tokens": 128000,
"mode": "chat",
@ -3010,7 +3010,7 @@
"input_cost_per_token": 5e-06,
"output_cost_per_token": 2.5e-05,
"litellm_provider": "azure_ai",
"max_input_tokens": 200000,
"max_input_tokens": 1000000,
"max_output_tokens": 128000,
"max_tokens": 128000,
"mode": "chat",
@ -45890,6 +45890,7 @@
"supports_native_streaming": true,
"supports_parallel_function_calling": true,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_response_schema": true,
"supports_system_messages": true,
"supports_tool_choice": true,
@ -45915,6 +45916,7 @@
"supports_native_streaming": true,
"supports_parallel_function_calling": true,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_response_schema": true,
"supports_system_messages": true,
"supports_tool_choice": true,
@ -45940,6 +45942,7 @@
"supports_native_streaming": true,
"supports_parallel_function_calling": true,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_response_schema": true,
"supports_system_messages": true,
"supports_tool_choice": true,
@ -45999,6 +46002,7 @@
"supports_native_streaming": true,
"supports_parallel_function_calling": true,
"supports_prompt_caching": true,
"supports_reasoning": true,
"supports_response_schema": true,
"supports_system_messages": true,
"supports_tool_choice": true,

9
litellm/proxy/_experimental/mcp_server/auth/litellm_auth_handler.py
View file

@ -1,12 +1,9 @@
from typing import TYPE_CHECKING, Dict, List, Optional
from typing import Dict, List, Optional
from mcp.server.auth.middleware.bearer_auth import AuthenticatedUser
from litellm.proxy._types import UserAPIKeyAuth
if TYPE_CHECKING:
from opentelemetry.trace import SpanContext
class MCPAuthenticatedUser(AuthenticatedUser):
"""
@ -19,8 +16,6 @@ class MCPAuthenticatedUser(AuthenticatedUser):
4. Server-specific authentication headers
5. OAuth2 headers
6. Raw headers - allows forwarding specific headers to the MCP server, specified by the admin.
7. Transport span context - the tracing span of the HTTP request carrying the current
message, which a stateful session's message handler cannot read from its own task.
"""
def __init__(
@ -33,7 +28,6 @@ class MCPAuthenticatedUser(AuthenticatedUser):
mcp_protocol_version: Optional[str] = None,
raw_headers: Optional[Dict[str, str]] = None,
client_ip: Optional[str] = None,
transport_span_context: Optional["SpanContext"] = None,
):
self.user_api_key_auth = user_api_key_auth
self.mcp_auth_header = mcp_auth_header
@ -43,4 +37,3 @@ class MCPAuthenticatedUser(AuthenticatedUser):
self.oauth2_headers = oauth2_headers
self.raw_headers = raw_headers
self.client_ip = client_ip
self.transport_span_context = transport_span_context

26
litellm/proxy/_experimental/mcp_server/db.py
View file

@ -9,10 +9,7 @@ from litellm._logging import verbose_proxy_logger
from litellm._uuid import uuid
from litellm.constants import MCP_PER_USER_TOKEN_EXPIRY_BUFFER_SECONDS
from litellm.llms.custom_httpx.http_handler import get_async_httpx_client
from litellm.proxy._experimental.mcp_server.auth.token_endpoint_auth import (
build_token_endpoint_client_auth,
normalize_token_endpoint_auth_method,
)
from litellm.proxy._experimental.mcp_server.oauth_utils import build_upstream_oauth2_token_request
from litellm.proxy._types import (
LiteLLM_MCPServerTable,
LiteLLM_ObjectPermissionTable,
@ -1248,11 +1245,12 @@ def _decrypted_credential_field(creds: Dict[str, object], field: str) -> object:
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).
spec_path for OpenAPI servers, plus the RFC 8707 upstream_resource sent on the authorize and
token legs), 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 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
@ -1278,6 +1276,7 @@ def mcp_oauth_token_identity(server: object) -> tuple[object, ...]:
_decrypted_credential_field(creds_dict, "client_id"),
_decrypted_credential_field(creds_dict, "client_secret"),
creds_dict.get("scopes"),
creds_dict.get("upstream_resource"),
)
@ -1367,20 +1366,21 @@ async def refresh_user_oauth_token(
return None
try:
client_auth = build_token_endpoint_client_auth(
auth_method=normalize_token_endpoint_auth_method(getattr(server, "token_endpoint_auth_method", None)),
token_request = build_upstream_oauth2_token_request(
server,
auth_method=getattr(server, "token_endpoint_auth_method", None),
client_id=client_id,
client_secret=client_secret,
)
token_data: Dict[str, str] = {
"grant_type": "refresh_token",
"refresh_token": refresh_token,
**client_auth.body,
**token_request.body,
}
async_client = get_async_httpx_client(llm_provider=httpxSpecialProvider.Oauth2Check)
response = await async_client.post(
token_url,
headers={"Accept": "application/json", **client_auth.headers},
headers={"Accept": "application/json", **token_request.headers},
data=token_data,
)
response.raise_for_status()

19
litellm/proxy/_experimental/mcp_server/discoverable_endpoints.py
View file

@ -21,7 +21,6 @@ from litellm.llms.custom_httpx.http_handler import (
)
from litellm.proxy._experimental.mcp_server.auth.token_endpoint_auth import (
TokenEndpointAuthConfigError,
build_token_endpoint_client_auth,
normalize_token_endpoint_auth_method,
)
from litellm.types.mcp_server.mcp_server_manager import MCPTokenEndpointAuthMethod
@ -54,7 +53,9 @@ from litellm.proxy._experimental.mcp_server.gateway_dcr_flow import (
)
from litellm.proxy._experimental.mcp_server.oauth_utils import (
TOKEN_NO_CACHE_HEADERS,
build_upstream_oauth2_token_request,
get_request_base_url,
resolve_upstream_resource,
validate_trusted_redirect_uri,
well_known_root_suffix,
)
@ -726,6 +727,7 @@ def _redirect_to_upstream_authorize(
to the upstream authorize endpoint verbatim, no relay state cookie is set, and the upstream
enforces its own registered redirect binding for the client."""
scope_value = scope or (" ".join(mcp_server.scopes) if mcp_server.scopes else None)
upstream_resource = resolve_upstream_resource(mcp_server)
passthrough_params = {
"client_id": client_id,
"redirect_uri": redirect_uri,
@ -734,6 +736,7 @@ def _redirect_to_upstream_authorize(
"code_challenge": code_challenge,
"code_challenge_method": code_challenge_method,
**({"scope": scope_value} if scope_value else {}),
**({"resource": upstream_resource} if upstream_resource else {}),
}
parsed_auth_url = urlparse(mcp_server.authorization_url or "")
merged_params = {**dict(parse_qsl(parsed_auth_url.query)), **passthrough_params}
@ -842,6 +845,10 @@ async def authorize_with_server(
if code_challenge_method:
params["code_challenge_method"] = code_challenge_method
upstream_resource = resolve_upstream_resource(mcp_server)
if upstream_resource:
params["resource"] = upstream_resource
parsed_auth_url = urlparse(mcp_server.authorization_url)
existing_params = dict(parse_qsl(parsed_auth_url.query))
existing_params.update(params)
@ -902,7 +909,8 @@ async def exchange_token_with_server(
else (client_token_endpoint_auth_method or mcp_server.token_endpoint_auth_method)
)
try:
client_auth = build_token_endpoint_client_auth(
token_request = build_upstream_oauth2_token_request(
mcp_server,
auth_method=resolved_auth_method,
client_id=resolved_client_id,
client_secret=resolved_client_secret,
@ -941,7 +949,7 @@ async def exchange_token_with_server(
token_data: dict = {
"grant_type": "refresh_token",
"refresh_token": upstream_refresh_token,
**client_auth.body,
**token_request.body,
}
refresh_request_scope = scope or bridge_upstream_scope
if refresh_request_scope:
@ -980,7 +988,7 @@ async def exchange_token_with_server(
"grant_type": "authorization_code",
"code": code,
"redirect_uri": resolved_redirect_uri,
**client_auth.body,
**token_request.body,
}
if code_verifier:
token_data["code_verifier"] = code_verifier
@ -991,11 +999,12 @@ async def exchange_token_with_server(
if not isinstance(prepared, _BridgeMintReady):
return _bridge_mint_error_response(prepared)
bridge_mint_ready = prepared
async_client = get_async_httpx_client(llm_provider=httpxSpecialProvider.Oauth2Check)
try:
response = await async_client.post(
mcp_server.token_url,
headers={"Accept": "application/json", **client_auth.headers},
headers={"Accept": "application/json", **token_request.headers},
data=token_data,
)
if response is not None:

17
litellm/proxy/_experimental/mcp_server/faults/classify.py
View file

@ -63,18 +63,21 @@ def _classify_oauth_error_code(
) -> UpstreamOAuthFault:
"""Blame assignment for a contract-conformant OAuth error code, shared by the token and DCR
classifiers. Codes by which the upstream blames itself keep that blame; ``invalid_target`` is a
gateway capability gap (RFC 8707 resource indicators, LIT-4339) no matter whose credentials were
presented; credential-indicting codes follow the credential source; everything else, including
codes we do not recognize, is the caller's to act on. The upstream's HTTP status is deliberately
never consulted: status derives from this classification at render time, which is what keeps
status and code from contradicting each other."""
gateway configuration gap (the RFC 8707 resource indicator this server sends, or fails to send)
no matter whose credentials were presented; credential-indicting codes follow the credential
source; everything else, including codes we do not recognize, is the caller's to act on. The
upstream's HTTP status is deliberately never consulted: status derives from this classification
at render time, which is what keeps status and code from contradicting each other."""
if code == "server_error" or code == "temporarily_unavailable":
return UpstreamReportedFault(code=code)
if code in GATEWAY_CAPABILITY_CODES:
verbose_logger.warning(
"MCP server %s: the upstream authorization server rejected the request with "
"invalid_target; it may require RFC 8707 resource indicators, which the gateway "
"does not send yet (tracked as LIT-4339)",
"invalid_target, meaning it did not accept the RFC 8707 resource indicator for this "
"request. Set upstream_resource on this server to the exact resource identifier the "
"authorization server expects (or to 'auto' to send the server's own canonical url); "
"if it is already set and the authorization server does not support resource "
"indicators, unset it and express the target audience through scopes instead",
log_context,
)
return GatewayRejected(code=code)

6
litellm/proxy/_experimental/mcp_server/faults/render_oauth.py
View file

@ -16,8 +16,10 @@ from litellm.proxy._experimental.mcp_server.oauth_utils import TOKEN_NO_CACHE_HE
def _gateway_rejected_description(code: str) -> str:
if code == "invalid_target":
return (
"the upstream authorization server rejected the request (invalid_target); "
"it may require RFC 8707 resource indicators, which the gateway does not send yet"
"the upstream authorization server rejected the request (invalid_target); it did not "
"accept this server's RFC 8707 resource indicator. Set upstream_resource on the MCP "
"server to the resource identifier the authorization server expects, or unset it if "
"that authorization server does not support resource indicators"
)
return (
f"the upstream authorization server rejected the gateway's configured client credentials "

7
litellm/proxy/_experimental/mcp_server/faults/types.py
View file

@ -25,9 +25,10 @@ gateway presented its own stored credentials, these are gateway-side faults the
when the caller supplied the credentials, they are the caller's to fix."""
GATEWAY_CAPABILITY_CODES: frozenset[str] = frozenset({"invalid_target"})
"""Codes that indict a gateway capability regardless of whose credentials were presented:
``invalid_target`` means the upstream wants RFC 8707 resource indicators, which the gateway does not
send yet (LIT-4339). Never the caller's fault."""
"""Codes that indict gateway configuration regardless of whose credentials were presented:
``invalid_target`` means the upstream did not accept the RFC 8707 resource indicator the server
sent, or requires one it was not configured to send (``upstream_resource``). Never the caller's
fault."""
UPSTREAM_FAULT_CODES: frozenset[str] = frozenset({"server_error", "temporarily_unavailable"})
"""Codes by which the upstream blames itself. Relaying them as caller faults would invert blame, so

38
litellm/proxy/_experimental/mcp_server/mcp_server_manager.py
View file

@ -71,6 +71,7 @@ from litellm.proxy._experimental.mcp_server.oauth2_token_cache import (
)
from litellm.proxy._experimental.mcp_server.oauth_utils import (
_redact_mcp_resource_url,
canonicalize_url_identity,
)
from litellm.proxy._experimental.mcp_server.outbound_credentials import (
Error,
@ -124,7 +125,6 @@ from litellm.proxy._experimental.mcp_server.utils import (
normalize_server_name,
parse_admin_env_vars,
split_server_prefix_from_name,
strip_known_server_prefix,
validate_mcp_server_name,
)
from litellm.proxy._types import (
@ -265,19 +265,11 @@ def _endpoints_yield_to_issuer(
def _normalized_authorize_endpoint(url: str) -> str:
"""Compare authorize endpoints on scheme, host, and path only. The default port is elided and
the host is lowercased so ``https://IDP.example.com:443/authorize/`` and
``https://idp.example.com/authorize`` are the same identity; query and trailing slash are not."""
parsed = urlparse(url)
scheme = parsed.scheme.lower()
host = (parsed.hostname or "").lower()
default_port = {"https": 443, "http": 80}.get(scheme)
try:
port = parsed.port
except ValueError:
port = None
authority = host if port is None or port == default_port else f"{host}:{port}"
return f"{scheme}://{authority}{parsed.path.rstrip('/')}"
"""Compare authorize endpoints / issuers on scheme, host, and path only, through the shared URL
canonicalizer: the default port is elided and the host is lowercased so
``https://IDP.example.com:443/authorize/`` and ``https://idp.example.com/authorize`` are the same
identity, while query, fragment and a trailing slash are dropped."""
return canonicalize_url_identity(url)
def _issuer_matches(claimed_issuer: object, configured_issuer: str) -> bool:
@ -1542,6 +1534,7 @@ class MCPServerManager:
"subject_token_type",
DEFAULT_SUBJECT_TOKEN_TYPE,
),
upstream_resource=server_config.get("upstream_resource", None),
# ID-JAG fields
id_jag_resource_token_endpoint=server_config.get("id_jag_resource_token_endpoint", None),
id_jag_resource=server_config.get("id_jag_resource", None),
@ -2041,6 +2034,7 @@ class MCPServerManager:
subject_token_type=mcp_server.subject_token_type
or (credentials_dict.get("subject_token_type") if credentials_dict else None)
or DEFAULT_SUBJECT_TOKEN_TYPE,
upstream_resource=(credentials_dict.get("upstream_resource") if credentials_dict else None),
# ID-JAG fields — read from credentials JSON blob
id_jag_resource_token_endpoint=(
credentials_dict.get("id_jag_resource_token_endpoint") if credentials_dict else None
@ -2516,6 +2510,13 @@ class MCPServerManager:
the given toolsets. Results are cached via ``user_api_key_cache`` (a
Redis-backed ``DualCache`` in production) so that cache entries are
shared across workers and cold-cache DB hits are minimised.
A row names a tool on the server identified by ``server_id``, so the
stored name is the tool's own name and is used as written. It is never
reduced by the server's wire prefix: that prefix is added on the way out
and is not part of any tool's identity, so treating a leading segment as
one silently renames the tool when a native name happens to begin with
it (``greyhound_internal_events`` on a server prefixed ``greyhound``).
"""
from litellm.proxy._experimental.mcp_server.toolset_db import list_mcp_toolsets
from litellm.proxy.proxy_server import prisma_client, user_api_key_cache
@ -2533,12 +2534,9 @@ class MCPServerManager:
tool_permissions: dict[str, list[str]] = {}
for toolset in toolsets:
for tool in toolset.tools:
raw_name = tool["tool_name"]
server = self.get_mcp_server_by_id(tool["server_id"])
unprefixed = strip_known_server_prefix(raw_name, server)
tool_permissions.setdefault(tool["server_id"], [])
if unprefixed not in tool_permissions[tool["server_id"]]:
tool_permissions[tool["server_id"]].append(unprefixed)
allowed_names = tool_permissions.setdefault(tool["server_id"], [])
if tool["tool_name"] not in allowed_names:
allowed_names.append(tool["tool_name"])
await user_api_key_cache.async_set_cache(
key=cache_key,
value=tool_permissions,

68
litellm/proxy/_experimental/mcp_server/oauth2_token_cache.py
View file

@ -6,6 +6,7 @@ with ``client_id``, ``client_secret``, and ``token_url``.
"""
import asyncio
import hashlib
from typing import TYPE_CHECKING, Dict, Optional, Tuple, Union
import httpx
@ -26,8 +27,9 @@ from litellm.proxy.common_utils.encrypt_decrypt_utils import (
decrypt_value_helper,
encrypt_value_helper,
)
from litellm.proxy._experimental.mcp_server.auth.token_endpoint_auth import (
build_token_endpoint_client_auth,
from litellm.proxy._experimental.mcp_server.oauth_utils import (
build_upstream_oauth2_token_request,
resolve_upstream_resource,
)
from litellm.types.llms.custom_http import httpxSpecialProvider
@ -37,10 +39,18 @@ if TYPE_CHECKING:
class MCPOAuth2TokenCache(InMemoryCache):
"""
In-memory cache for OAuth2 client_credentials tokens, keyed by server_id.
In-memory cache for OAuth2 client_credentials tokens, keyed by the identity of the token
request rather than by server_id alone.
A minted token is only reusable for the exact request that produced it. Keying on server_id
alone served a token minted under the previous configuration whenever any of those inputs
changed, so editing scopes, rotating the client secret, or setting ``upstream_resource``
silently kept handing out a token carrying the old scopes or audience until it expired. The
identity below covers every input ``_fetch_token`` puts on the wire, so a change to any of
them misses the cache and mints afresh.
Inherits from ``InMemoryCache`` for TTL-based storage and eviction.
Adds per-server ``asyncio.Lock`` to prevent duplicate concurrent fetches.
Adds a per-identity ``asyncio.Lock`` to prevent duplicate concurrent fetches.
"""
def __init__(self) -> None:
@ -50,8 +60,25 @@ class MCPOAuth2TokenCache(InMemoryCache):
)
self._locks: Dict[str, asyncio.Lock] = {}
def _get_lock(self, server_id: str) -> asyncio.Lock:
return self._locks.setdefault(server_id, asyncio.Lock())
@staticmethod
def _token_identity(server: "MCPServer") -> str:
"""Cache key for the token this server's config would mint, prefixed by server_id so a
single server's entries stay greppable and invalidatable. The secret is hashed with the
rest of the identity rather than stored in a key."""
material = "\x00".join(
(
server.token_url or "",
server.client_id or "",
server.client_secret or "",
" ".join(server.scopes or ()),
resolve_upstream_resource(server) or "",
server.token_endpoint_auth_method or "",
)
)
return f"{server.server_id}:{hashlib.sha256(material.encode()).hexdigest()}"
def _get_lock(self, identity: str) -> asyncio.Lock:
return self._locks.setdefault(identity, asyncio.Lock())
@staticmethod
def _has_client_credentials_config(server: "MCPServer") -> bool:
@ -67,21 +94,21 @@ class MCPOAuth2TokenCache(InMemoryCache):
if not self._has_client_credentials_config(server):
return None
server_id = server.server_id
identity = self._token_identity(server)
# Fast path — cached token is still valid
cached = self.get_cache(server_id)
cached = self.get_cache(identity)
if cached is not None:
return cached
# Slow path — acquire per-server lock then double-check
async with self._get_lock(server_id):
cached = self.get_cache(server_id)
# Slow path — acquire per-identity lock then double-check
async with self._get_lock(identity):
cached = self.get_cache(identity)
if cached is not None:
return cached
token, ttl = await self._fetch_token(server)
self.set_cache(server_id, token, ttl=ttl)
self.set_cache(identity, token, ttl=ttl)
return token
async def _fetch_token(self, server: "MCPServer") -> Tuple[str, int]:
@ -100,14 +127,15 @@ class MCPOAuth2TokenCache(InMemoryCache):
f"token_url={bool(server.token_url)}"
)
client_auth = build_token_endpoint_client_auth(
token_request = build_upstream_oauth2_token_request(
server,
auth_method=server.token_endpoint_auth_method,
client_id=server.client_id,
client_secret=server.client_secret,
)
data: Dict[str, str] = {
"grant_type": "client_credentials",
**client_auth.body,
**token_request.body,
}
if server.scopes:
data["scope"] = " ".join(server.scopes)
@ -117,7 +145,7 @@ class MCPOAuth2TokenCache(InMemoryCache):
server.server_id,
)
post_kwargs = {"data": data, **({"headers": client_auth.headers} if client_auth.headers else {})}
post_kwargs = {"data": data, **({"headers": token_request.headers} if token_request.headers else {})}
try:
response = await client.post(server.token_url, **post_kwargs)
response.raise_for_status()
@ -159,8 +187,14 @@ class MCPOAuth2TokenCache(InMemoryCache):
return access_token, ttl
def invalidate(self, server_id: str) -> None:
"""Remove a cached token (e.g. after a 401)."""
self.delete_cache(server_id)
"""Remove every cached token for a server (e.g. after a 401).
Entries are keyed by token identity, so one server can hold more than one entry across a
config change; a 401 invalidates all of them rather than only the current configuration's.
"""
prefix = f"{server_id}:"
for key in [k for k in self.cache_dict if isinstance(k, str) and k.startswith(prefix)]:
self.delete_cache(key)
mcp_oauth2_token_cache = MCPOAuth2TokenCache()

123
litellm/proxy/_experimental/mcp_server/oauth_utils.py
View file

@ -3,14 +3,22 @@
import os
from ipaddress import ip_address
from typing import Any, Dict, List, NoReturn, Optional
from typing import TYPE_CHECKING, Any, Dict, List, NoReturn, Optional
from urllib.parse import ParseResult, urlparse, urlsplit, urlunparse, urlunsplit
from fastapi import HTTPException, Request
from litellm._logging import verbose_logger
from litellm.proxy._experimental.mcp_server.auth.token_endpoint_auth import (
TokenEndpointClientAuth,
build_token_endpoint_client_auth,
normalize_token_endpoint_auth_method,
)
from litellm.proxy.auth.ip_address_utils import IPAddressUtils
if TYPE_CHECKING:
from litellm.types.mcp_server.mcp_server_manager import MCPServer
# RFC 6749 §5.1 / OAuth 2.1 draft-15 §4.1.3: token-endpoint responses
# must not be cached — both success and error bodies may reveal secrets.
TOKEN_NO_CACHE_HEADERS = {"Cache-Control": "no-store", "Pragma": "no-cache"}
@ -21,6 +29,10 @@ TOKEN_NO_CACHE_HEADERS = {"Cache-Control": "no-store", "Pragma": "no-cache"}
# explicit port, which would otherwise break a literal netloc compare).
_DEFAULT_PORTS = {"http": 80, "https": 443}
# Sentinel ``upstream_resource`` value meaning "derive the RFC 8707 resource identifier from the
# server's own url". RFC 8707 requires an absolute URI, so this can never be a real resource value.
UPSTREAM_RESOURCE_AUTO = "auto"
# Env var for ops to allowlist additional redirect_uri origins beyond
# same-origin + loopback — needed for first-party OAuth clients hosted
# on sister domains (e.g. a web app on app.example.com registering as
@ -574,3 +586,112 @@ def validate_trusted_redirect_uri(request: Request, redirect_uri: str) -> None:
if _trusted_redirect_uri_is_allowed(parsed, redirect_netloc, proxy_base):
return
_raise_trusted_redirect_uri_rejected(request, redirect_uri, parsed, redirect_netloc, proxy_base)
def canonicalize_url_identity(url: str) -> str:
"""Normalize a URL to a comparable identity: lowercase scheme and host, drop the scheme's default
port, and strip userinfo, params, query, fragment and a trailing slash while keeping IPv6
brackets. The one URL-canonicalization primitive shared by the RFC 8707 resource emitter and the
RFC 8414 issuer/authorize-endpoint comparison, so the default-port and IPv6 rules cannot be
present in one and missing in the other. The netloc (not ``parsed.hostname``) carries the
authority so ``[::1]:8080`` survives with its brackets intact."""
parsed = urlparse(url)
scheme = parsed.scheme.lower()
netloc = _strip_default_port(scheme, parsed.netloc.rpartition("@")[2])
return urlunparse((scheme, netloc, parsed.path.rstrip("/"), "", "", ""))
def _canonical_resource_uri(url: str) -> str | None:
"""Canonicalize an upstream MCP server URL into an RFC 8707 resource identifier.
Keeps only the scheme, host, port and path, which is the shape the MCP authorization spec's
"Canonical Server URI" section describes and every one of its examples takes; the reference
implementation is ``mcp.shared.auth_utils.resource_url_from_server_url``, and this is the stricter
variant. The scheme and host are lowercased, the scheme's default port is dropped so
``https://host:443/mcp`` and ``https://host/mcp`` never present as two resources, and a trailing
slash is dropped so ``https://host/mcp/`` and ``https://host/mcp`` do not either.
Userinfo, query and fragment are dropped rather than carried. A transport URL routinely holds
credentials in exactly those components (``user:password@``, ``?api_key=``), while a resource
indicator names the resource and nothing else; this value is published somewhere the transport
URL never goes, into the authorization redirect the browser follows and into token request
bodies, so carrying them would disclose them to the authorization server, its logs, and browser
history. RFC 8707 forbids a fragment outright and says a resource SHOULD NOT carry a query. An
upstream whose identifier genuinely needs more than this is served by setting
``upstream_resource`` explicitly, which is passed through untouched.
Returns ``None`` when the URL is not absolute, which cannot yield a valid resource identifier.
"""
parsed = urlparse(url)
if not parsed.scheme or not parsed.netloc:
return None
return canonicalize_url_identity(url)
def resolve_upstream_resource(mcp_server: "MCPServer") -> str | None:
"""Resolve the RFC 8707 ``resource`` value this server's upstream OAuth legs must carry.
The MCP authorization spec requires an MCP client to send ``resource`` on both the
authorization request and every token request, naming the canonical URI of the MCP server the
token is for. Authorization server temperaments are irreconcilable and undetectable, so this
stays an explicit per-server opt-in: most SaaS providers ignore the parameter, some hard-reject
it and express audience through scopes instead, and strict or MCP-native ones refuse to mint a
correctly scoped token without it (``invalid_target``).
``None`` or blank omits the parameter, which is the default and preserves the behavior of every
server working today. ``"auto"`` derives the canonical URI from the server's own URL; it is not
an absolute URI, so RFC 8707 guarantees it can never collide with a real resource value. Any
other value is sent verbatim, because the identifier has to match what the authorization server
expects exactly and normalizing it could break that match.
Every upstream leg for a server resolves through this one function, so the authorize request
and the token requests cannot disagree; a token request naming a resource the authorization
request never asked for is itself an ``invalid_target`` under RFC 8707.
"""
configured = (mcp_server.upstream_resource or "").strip()
if not configured:
return None
if configured.lower() != UPSTREAM_RESOURCE_AUTO:
return configured
if not mcp_server.url:
verbose_logger.warning(
"MCP server %s sets upstream_resource=auto but has no url to derive a resource "
"identifier from; omitting the RFC 8707 resource parameter. Set upstream_resource to "
"the exact resource identifier the authorization server expects instead.",
mcp_server.server_id,
)
return None
canonical = _canonical_resource_uri(mcp_server.url)
if canonical is None:
verbose_logger.warning(
"MCP server %s sets upstream_resource=auto but its url is not an absolute URI, so no "
"RFC 8707 resource identifier could be derived; omitting the resource parameter",
mcp_server.server_id,
)
return canonical
def build_upstream_oauth2_token_request(
mcp_server: "MCPServer",
*,
auth_method: object,
client_id: str | None,
client_secret: str | None,
) -> TokenEndpointClientAuth:
"""Client auth plus the RFC 8707 ``resource`` for one upstream plain-OAuth2 token request.
Resolving both in one call is what stops a leg authenticating without naming the resource its
sibling legs named; the RFC 8693 legs (OBO, id_jag) carry ``audience`` and stay on
``build_token_endpoint_client_auth``. The client-auth inputs are passed in because a leg may
authenticate as the caller's own client rather than the server's; ``resource`` always comes from
the server, so no leg can choose or forget it.
"""
client_auth = build_token_endpoint_client_auth(
auth_method=normalize_token_endpoint_auth_method(auth_method),
client_id=client_id,
client_secret=client_secret,
)
resource = resolve_upstream_resource(mcp_server)
if not resource:
return client_auth
return TokenEndpointClientAuth(headers=client_auth.headers, body={**client_auth.body, "resource": resource})

2
litellm/proxy/_experimental/mcp_server/outbound_credentials/adapter.py
View file

@ -21,6 +21,7 @@ from typing_extensions import assert_never
from litellm.proxy._experimental.mcp_server.outbound_credentials.credential_provenance import (
classify_inbound_provenance,
)
from litellm.proxy._experimental.mcp_server.oauth_utils import resolve_upstream_resource
from litellm.proxy._experimental.mcp_server.outbound_credentials.types import (
ApiKeyConfig,
AuthorizationCodeConfig,
@ -151,6 +152,7 @@ def _client_credentials_spec(server: MCPServer, resource: str) -> ServerSpec:
token_url=server.token_url,
scopes=tuple(server.scopes or ()),
audience=server.audience,
upstream_resource=resolve_upstream_resource(server),
token_endpoint_auth_method=server.token_endpoint_auth_method,
),
)

9
litellm/proxy/_experimental/mcp_server/outbound_credentials/authz_code_refresher.py
View file

@ -17,8 +17,8 @@ from typing import TYPE_CHECKING, Protocol
from litellm._logging import verbose_logger
from litellm.proxy._experimental.mcp_server.auth.token_endpoint_auth import (
TokenEndpointAuthConfigError,
build_token_endpoint_client_auth,
)
from litellm.proxy._experimental.mcp_server.oauth_utils import build_upstream_oauth2_token_request
from litellm.proxy._experimental.mcp_server.outbound_credentials.oauth_token_store import (
OAuthToken,
)
@ -92,7 +92,8 @@ class AuthorizationCodeRefresher:
return None
try:
client_auth = build_token_endpoint_client_auth(
token_request = build_upstream_oauth2_token_request(
server,
auth_method=server.token_endpoint_auth_method,
client_id=server.client_id,
client_secret=server.client_secret,
@ -103,9 +104,9 @@ class AuthorizationCodeRefresher:
form = {
"grant_type": "refresh_token",
"refresh_token": token.refresh_token,
**client_auth.body,
**token_request.body,
}
body = await self._token_endpoint(server.token_url, form, client_auth.headers)
body = await self._token_endpoint(server.token_url, form, token_request.headers)
if body is None:
return None
access_token = body.get("access_token")

2
litellm/proxy/_experimental/mcp_server/outbound_credentials/client_credentials.py
View file

@ -292,6 +292,7 @@ def _prepare_grant(config: ClientCredentialsConfig) -> Result[_PreparedGrant, Cr
**client_auth.body,
**({"scope": " ".join(config.scopes)} if config.scopes else {}),
**({"audience": config.audience} if config.audience else {}),
**({"resource": config.upstream_resource} if config.upstream_resource else {}),
}
return Ok(
_PreparedGrant(
@ -313,6 +314,7 @@ def _identity_key(config: ClientCredentialsConfig) -> str:
config.token_endpoint_auth_method or "",
" ".join(config.scopes),
config.audience or "",
config.upstream_resource or "",
)
)
return hashlib.sha256(material.encode("utf-8")).hexdigest()

1
litellm/proxy/_experimental/mcp_server/outbound_credentials/types.py
View file

@ -199,6 +199,7 @@ class ClientCredentialsConfig(BaseModel):
token_url: str | None = None
scopes: tuple[str, ...] = ()
audience: str | None = None
upstream_resource: str | None = None
token_endpoint_auth_method: Literal["client_secret_post", "client_secret_basic"] | None = None

106
litellm/proxy/_experimental/mcp_server/server.py
View file

@ -15,7 +15,6 @@ import types
import uuid
from datetime import datetime
from typing import (
TYPE_CHECKING,
Any,
AsyncIterator,
Callable,
@ -107,9 +106,9 @@ _MAX_STATEFUL_SESSIONS_PER_OWNER = 100
# prevents an authenticated client from forcing the proxy to buffer an
# arbitrarily large body just to make a routing decision.
_MCP_ROUTING_PEEK_MAX_BYTES = 4096
if TYPE_CHECKING:
from opentelemetry.trace import SpanContext
# ASGI scope key holding the tracing span of the request carrying an MCP
# message, written on the request task and read back by the message handler.
_MCP_TRANSPORT_SPAN_SCOPE_KEY = "litellm_otel_transport_span"
def _invalidate_byok_cred_cache(user_id: str, server_id: str) -> None:
@ -294,52 +293,78 @@ def _otel_reset_mcp_trace_carrier(token: object) -> None:
return
def _otel_request_transport_span_context() -> Optional["SpanContext"]:
"""The tracing span of the HTTP request being handled, as a portable value.
def _otel_publish_transport_span_on_scope(scope: Scope) -> None:
"""Record this request's tracing span on its own ASGI scope.
Resolved on the ASGI request task, where the proxy's server span is anchored,
and carried to the MCP message handler on the authenticated-user object. A
stateful streamable-HTTP session handles every message on the task spawned by
its ``initialize`` POST, so the handler's own task cannot see later requests'
spans; this is the same reason per-request auth is carried across rather than
read from a ContextVar. Lazily imported so opentelemetry stays an optional
dependency; returns ``None`` when otel_v2 is unavailable or no request span is
and read back by the MCP message handler through ``req_ctx.request`` — the
``Request`` the transport attaches to each message. A stateful streamable-HTTP
session handles every message on the task spawned by its ``initialize`` POST, so
the handler's own task cannot see later requests' spans.
The scope, not the shared session auth context: a JSON-RPC *response* POST
deliberately skips the per-session lock (it can arrive while the tool call that
awaits it is still in flight), so a field on that shared object would be
overwritten mid-call and the tool call would attribute itself to the response's
request. A scope belongs to exactly one request and dies with it, which also
keeps a finished span from being retained by an idle session.
The live span, not just its context: a failed tool call stamps ``error.*`` on it,
which needs a span still open for writes. Lazily imported so opentelemetry stays
an optional dependency; a no-op when otel_v2 is unavailable or no request span is
anchored."""
try:
from litellm.integrations.otel.plumbing.context import (
request_root_span_context,
request_root_span,
)
return request_root_span_context()
span = request_root_span()
except ImportError:
return
if span is not None:
scope[_MCP_TRANSPORT_SPAN_SCOPE_KEY] = span
def _otel_transport_span_from_message(req_ctx: object) -> object:
"""The tracing span of the HTTP request that carried this MCP message.
Read off that request's ASGI scope, reached through the ``Request`` the
streamable-HTTP transport attaches to each message, so it is this message's
transport and not whichever request happens to have touched the session last.
Returns whatever the scope holds; the otel plumbing validates it."""
request = getattr(req_ctx, "request", None)
scope = getattr(request, "scope", None)
if not isinstance(scope, Mapping):
return None
return scope.get(_MCP_TRANSPORT_SPAN_SCOPE_KEY)
def _otel_set_mcp_transport_span_context(span_context: Optional["SpanContext"]) -> object:
"""Publish the current message's transport span for the otel_v2 MCP span and
return a reset token, or ``None`` when otel_v2 is unavailable."""
if span_context is None:
def _otel_set_mcp_transport_span(span: object) -> object:
"""Publish the current message's transport span, which the otel_v2 MCP span
attaches to and a failed tool call stamps its error on. Returns a reset token,
or ``None`` when otel_v2 is unavailable."""
if span is None:
return None
try:
from litellm.integrations.otel.plumbing.context import (
set_mcp_message_transport_span_context,
set_mcp_message_transport_span,
)
return set_mcp_message_transport_span_context(span_context)
return set_mcp_message_transport_span(span)
except ImportError:
return None
def _otel_reset_mcp_transport_span_context(token: object) -> None:
"""Paired with ``_otel_set_mcp_transport_span_context``."""
def _otel_reset_mcp_transport_span(token: object) -> None:
"""Paired with ``_otel_set_mcp_transport_span``."""
if token is None:
return
try:
from litellm.integrations.otel.plumbing.context import (
reset_mcp_message_transport_span_context,
reset_mcp_message_transport_span,
)
reset_mcp_message_transport_span_context(token)
reset_mcp_message_transport_span(token)
except ImportError:
return
@ -710,18 +735,6 @@ if MCP_AVAILABLE:
############### MCP Server Routes #######################
########################################################
def _current_transport_span_context() -> Optional["SpanContext"]:
"""The transport span of the HTTP request carrying the message being handled.
Published by the ASGI request task onto the authenticated-user object, because
a stateful session's message handler runs on the task spawned by that session's
``initialize`` POST and so cannot read later requests' spans from its own task.
"""
auth_user = auth_context_var.get()
if not isinstance(auth_user, MCPAuthenticatedUser):
auth_user = _recover_auth_from_session()
return auth_user.transport_span_context if auth_user is not None else None
@server.list_tools()
async def handle_list_tools() -> "ListToolsResult | List[Tool]":
"""
@ -742,7 +755,7 @@ if MCP_AVAILABLE:
try:
_trace_token = _otel_set_mcp_trace_carrier(_mcp_meta_trace_carrier(req_ctx))
_transport_token = _otel_set_mcp_transport_span_context(_current_transport_span_context())
_transport_token = _otel_set_mcp_transport_span(_otel_transport_span_from_message(req_ctx))
# Get user authentication from context variable
(
user_api_key_auth,
@ -798,7 +811,7 @@ if MCP_AVAILABLE:
# This prevents the HTTP stream from failing and allows the client to get a response
return []
finally:
_otel_reset_mcp_transport_span_context(_transport_token)
_otel_reset_mcp_transport_span(_transport_token)
_otel_reset_mcp_trace_carrier(_trace_token)
if _session_reset_token is not None:
active_mcp_session_var.reset(_session_reset_token)
@ -976,7 +989,7 @@ if MCP_AVAILABLE:
try:
_trace_token = _otel_set_mcp_trace_carrier(_mcp_meta_trace_carrier(req_ctx))
_transport_token = _otel_set_mcp_transport_span_context(_current_transport_span_context())
_transport_token = _otel_set_mcp_transport_span(_otel_transport_span_from_message(req_ctx))
# Validate arguments
(
user_api_key_auth,
@ -1115,7 +1128,7 @@ if MCP_AVAILABLE:
return response
finally:
_otel_reset_mcp_transport_span_context(_transport_token)
_otel_reset_mcp_transport_span(_transport_token)
_otel_reset_mcp_trace_carrier(_trace_token)
if _session_reset_token is not None:
active_mcp_session_var.reset(_session_reset_token)
@ -2717,7 +2730,7 @@ if MCP_AVAILABLE:
):
raise HTTPException(
status_code=403,
detail=f"User not allowed to call this tool. Allowed MCP servers: {allowed_mcp_servers}",
detail="User not allowed to call this tool.",
)
standard_logging_mcp_tool_call: StandardLoggingMCPToolCall = _get_standard_logging_mcp_tool_call(
@ -4282,6 +4295,7 @@ if MCP_AVAILABLE:
_increment_active_request_session(initialized_session_id)
async def _dispatch() -> None:
_otel_publish_transport_span_on_scope(scope)
auth_user = _set_or_update_auth_context(
user_api_key_auth=user_api_key_auth,
mcp_auth_header=mcp_auth_header,
@ -4293,7 +4307,6 @@ if MCP_AVAILABLE:
session_id=session_id if use_stateful else None,
touch_last_seen=(scope.get("method") or "").upper() != "DELETE",
copy_existing_session_auth_context=is_initialize,
transport_span_context=_otel_request_transport_span_context(),
)
local_send = send
if use_stateful and is_initialize:
@ -4519,7 +4532,6 @@ if MCP_AVAILABLE:
oauth2_headers: Optional[Dict[str, str]] = None,
raw_headers: Optional[Dict[str, str]] = None,
client_ip: Optional[str] = None,
transport_span_context: Optional["SpanContext"] = None,
) -> None:
auth_user.user_api_key_auth = user_api_key_auth
auth_user.mcp_auth_header = mcp_auth_header
@ -4528,7 +4540,6 @@ if MCP_AVAILABLE:
auth_user.oauth2_headers = oauth2_headers
auth_user.raw_headers = raw_headers
auth_user.client_ip = client_ip
auth_user.transport_span_context = transport_span_context
def set_auth_context(
user_api_key_auth: Optional[UserAPIKeyAuth],
@ -4538,7 +4549,6 @@ if MCP_AVAILABLE:
oauth2_headers: Optional[Dict[str, str]] = None,
raw_headers: Optional[Dict[str, str]] = None,
client_ip: Optional[str] = None,
transport_span_context: Optional["SpanContext"] = None,
) -> MCPAuthenticatedUser:
"""
Set the UserAPIKeyAuth in the auth context variable.
@ -4549,7 +4559,6 @@ if MCP_AVAILABLE:
mcp_servers: Optional list of server names and access groups to filter by
mcp_server_auth_headers: Optional dict of server-specific auth headers {server_alias: auth_value}
client_ip: Client IP address for MCP access control
transport_span_context: Tracing span of the HTTP request carrying this message
"""
auth_user = MCPAuthenticatedUser(
user_api_key_auth=user_api_key_auth,
@ -4559,7 +4568,6 @@ if MCP_AVAILABLE:
oauth2_headers=oauth2_headers,
raw_headers=raw_headers,
client_ip=client_ip,
transport_span_context=transport_span_context,
)
auth_context_var.set(auth_user)
return auth_user
@ -4575,7 +4583,6 @@ if MCP_AVAILABLE:
session_id: Optional[str] = None,
touch_last_seen: bool = True,
copy_existing_session_auth_context: bool = False,
transport_span_context: Optional["SpanContext"] = None,
) -> MCPAuthenticatedUser:
auth_user = _stateful_session_auth_contexts.get(session_id) if session_id else None
if auth_user is not None and session_id is not None:
@ -4590,7 +4597,6 @@ if MCP_AVAILABLE:
oauth2_headers=oauth2_headers,
raw_headers=raw_headers,
client_ip=client_ip,
transport_span_context=transport_span_context,
)
_update_auth_context(
auth_user=auth_user,
@ -4601,7 +4607,6 @@ if MCP_AVAILABLE:
oauth2_headers=oauth2_headers,
raw_headers=raw_headers,
client_ip=client_ip,
transport_span_context=transport_span_context,
)
auth_context_var.set(auth_user)
return auth_user
@ -4613,7 +4618,6 @@ if MCP_AVAILABLE:
oauth2_headers=oauth2_headers,
raw_headers=raw_headers,
client_ip=client_ip,
transport_span_context=transport_span_context,
)
def _wrap_send_with_stateful_session_auth_context(

2
litellm/proxy/_experimental/out/404.html
View file

File diff suppressed because one or more lines are too long

2
litellm/proxy/_experimental/out/404/index.html
View file

File diff suppressed because one or more lines are too long

8
litellm/proxy/_experimental/out/__next.!KGRhc2hib2FyZCk.__PAGE__.txt
View file

@ -1,9 +1,9 @@
1:"$Sreact.fragment"
2:I[347257,["/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js"],"ClientPageRoot"]
3:I[871135,["/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js","/litellm-asset-prefix/_next/static/chunks/0mqbd99.ej13v.js","/litellm-asset-prefix/_next/static/chunks/16ufy1iyybswo.js","/litellm-asset-prefix/_next/static/chunks/0ayum-x.hkww~.js","/litellm-asset-prefix/_next/static/chunks/15jl-1gcakfwa.js","/litellm-asset-prefix/_next/static/chunks/003_1s9xbht43.js","/litellm-asset-prefix/_next/static/chunks/0u.r3vzo30ofk.js","/litellm-asset-prefix/_next/static/chunks/0g4qcx-c9gsxn.js","/litellm-asset-prefix/_next/static/chunks/14a-un1blorp~.js","/litellm-asset-prefix/_next/static/chunks/02_q4881cz6h~.js","/litellm-asset-prefix/_next/static/chunks/0lg.6rbfsd-l9.js","/litellm-asset-prefix/_next/static/chunks/0wdlbe750tuzr.js","/litellm-asset-prefix/_next/static/chunks/043q3g5-5-aju.js","/litellm-asset-prefix/_next/static/chunks/0nht59ws0elww.js","/litellm-asset-prefix/_next/static/chunks/03zkt5iyjiqcz.js","/litellm-asset-prefix/_next/static/chunks/0l02mpo6za6ie.js","/litellm-asset-prefix/_next/static/chunks/1667t2pcy0iqm.js","/litellm-asset-prefix/_next/static/chunks/0-_m4km7b1~oe.js","/litellm-asset-prefix/_next/static/chunks/0sx3mu2_l9g_y.js","/litellm-asset-prefix/_next/static/chunks/0sqw622fcvsv4.js","/litellm-asset-prefix/_next/static/chunks/0t8el_ijoskx..js","/litellm-asset-prefix/_next/static/chunks/0el08tticy_20.js","/litellm-asset-prefix/_next/static/chunks/0a.ljputcx8g5.js","/litellm-asset-prefix/_next/static/chunks/02-u6qtmsnqn0.js","/litellm-asset-prefix/_next/static/chunks/076.vm.7w-x2..js","/litellm-asset-prefix/_next/static/chunks/069dx5~5osue0.js","/litellm-asset-prefix/_next/static/chunks/0onea0n77pqw1.js","/litellm-asset-prefix/_next/static/chunks/16jfov1k2wrj0.js","/litellm-asset-prefix/_next/static/chunks/0l7em-5kjv49e.js","/litellm-asset-prefix/_next/static/chunks/0mh1wnrvmv_y7.js","/litellm-asset-prefix/_next/static/chunks/01reddhq423_f.js","/litellm-asset-prefix/_next/static/chunks/0h80lrrstjswl.js","/litellm-asset-prefix/_next/static/chunks/0cmepm.jkel-i.js","/litellm-asset-prefix/_next/static/chunks/0q6~n4y84cejn.js","/litellm-asset-prefix/_next/static/chunks/04jv9e6~9vi.l.js","/litellm-asset-prefix/_next/static/chunks/0d_sm.._5mw-p.js","/litellm-asset-prefix/_next/static/chunks/0elr0ye86.44-.js"],"default"]
6:I[897367,["/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js"],"OutletBoundary"]
2:I[347257,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"ClientPageRoot"]
3:I[871135,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js","/litellm-asset-prefix/_next/static/chunks/1ioy8obpggx93.js","/litellm-asset-prefix/_next/static/chunks/3_3dj4vdy-3xy.js","/litellm-asset-prefix/_next/static/chunks/1zr7rrk4wkmju.js","/litellm-asset-prefix/_next/static/chunks/2cngn5bal3278.js","/litellm-asset-prefix/_next/static/chunks/2up3bks93iqds.js","/litellm-asset-prefix/_next/static/chunks/1l2mgm5v3tjci.js","/litellm-asset-prefix/_next/static/chunks/0zduf1gntl_f8.js","/litellm-asset-prefix/_next/static/chunks/0g_w4tf2inv3i.js","/litellm-asset-prefix/_next/static/chunks/1a0bgy7kzrj91.js","/litellm-asset-prefix/_next/static/chunks/0dbvgsc7ha049.js","/litellm-asset-prefix/_next/static/chunks/17-6zku8f68gf.js","/litellm-asset-prefix/_next/static/chunks/0f5fel02jwglw.js","/litellm-asset-prefix/_next/static/chunks/1vquuz09jxl5_.js","/litellm-asset-prefix/_next/static/chunks/1iakmimqrlpn0.js","/litellm-asset-prefix/_next/static/chunks/2n26sdz53rm0a.js","/litellm-asset-prefix/_next/static/chunks/11khk745tfruy.js","/litellm-asset-prefix/_next/static/chunks/00g6xfr4yow7h.js","/litellm-asset-prefix/_next/static/chunks/0-k_4_s7m108w.js","/litellm-asset-prefix/_next/static/chunks/2kcxwg1mpncp6.js","/litellm-asset-prefix/_next/static/chunks/3drq2_k-jeio2.js","/litellm-asset-prefix/_next/static/chunks/395_vbpmrlvpu.js","/litellm-asset-prefix/_next/static/chunks/2hu1vyy-5pv13.js","/litellm-asset-prefix/_next/static/chunks/2l25bmiiw9ixp.js","/litellm-asset-prefix/_next/static/chunks/1uz3jt-tj9lkf.js","/litellm-asset-prefix/_next/static/chunks/199uwr871eene.js","/litellm-asset-prefix/_next/static/chunks/2s_ce-opzrkzr.js","/litellm-asset-prefix/_next/static/chunks/2ptdxz8qnchh_.js","/litellm-asset-prefix/_next/static/chunks/17nqbxvhztf3k.js","/litellm-asset-prefix/_next/static/chunks/2c90xukbd3il6.js","/litellm-asset-prefix/_next/static/chunks/0kap_rdm2-lem.js","/litellm-asset-prefix/_next/static/chunks/09l_m9l1emin2.js","/litellm-asset-prefix/_next/static/chunks/1cea03gg5a_c7.js","/litellm-asset-prefix/_next/static/chunks/323l6h8s7ahat.js","/litellm-asset-prefix/_next/static/chunks/112n0hv3cc2rg.js","/litellm-asset-prefix/_next/static/chunks/3y674jhwchpcq.js","/litellm-asset-prefix/_next/static/chunks/105643dvf00hu.js","/litellm-asset-prefix/_next/static/chunks/12wsfsljxg4xv.js","/litellm-asset-prefix/_next/static/chunks/22iools_e0k44.js","/litellm-asset-prefix/_next/static/chunks/3f9uewf5w-e-p.js","/litellm-asset-prefix/_next/static/chunks/23-g73xaw3kap.js"],"default"]
6:I[897367,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"OutletBoundary"]
7:"$Sreact.suspense"
0:{"rsc":["$","$1","c",{"children":[["$","$L2",null,{"Component":"$3","serverProvidedParams":{"searchParams":{},"params":{},"promises":["$@4","$@5"]}}],[["$","script","script-0",{"src":"/litellm-asset-prefix/_next/static/chunks/1667t2pcy0iqm.js","async":true}],["$","script","script-1",{"src":"/litellm-asset-prefix/_next/static/chunks/0-_m4km7b1~oe.js","async":true}],["$","script","script-2",{"src":"/litellm-asset-prefix/_next/static/chunks/0sx3mu2_l9g_y.js","async":true}],["$","script","script-3",{"src":"/litellm-asset-prefix/_next/static/chunks/0sqw622fcvsv4.js","async":true}],["$","script","script-4",{"src":"/litellm-asset-prefix/_next/static/chunks/0t8el_ijoskx..js","async":true}],["$","script","script-5",{"src":"/litellm-asset-prefix/_next/static/chunks/0el08tticy_20.js","async":true}],["$","script","script-6",{"src":"/litellm-asset-prefix/_next/static/chunks/0a.ljputcx8g5.js","async":true}],["$","script","script-7",{"src":"/litellm-asset-prefix/_next/static/chunks/02-u6qtmsnqn0.js","async":true}],["$","script","script-8",{"src":"/litellm-asset-prefix/_next/static/chunks/076.vm.7w-x2..js","async":true}],["$","script","script-9",{"src":"/litellm-asset-prefix/_next/static/chunks/069dx5~5osue0.js","async":true}],["$","script","script-10",{"src":"/litellm-asset-prefix/_next/static/chunks/0onea0n77pqw1.js","async":true}],["$","script","script-11",{"src":"/litellm-asset-prefix/_next/static/chunks/16jfov1k2wrj0.js","async":true}],["$","script","script-12",{"src":"/litellm-asset-prefix/_next/static/chunks/0l7em-5kjv49e.js","async":true}],["$","script","script-13",{"src":"/litellm-asset-prefix/_next/static/chunks/0mh1wnrvmv_y7.js","async":true}],["$","script","script-14",{"src":"/litellm-asset-prefix/_next/static/chunks/01reddhq423_f.js","async":true}],["$","script","script-15",{"src":"/litellm-asset-prefix/_next/static/chunks/0h80lrrstjswl.js","async":true}],["$","script","script-16",{"src":"/litellm-asset-prefix/_next/static/chunks/0cmepm.jkel-i.js","async":true}],["$","script","script-17",{"src":"/litellm-asset-prefix/_next/static/chunks/0q6~n4y84cejn.js","async":true}],["$","script","script-18",{"src":"/litellm-asset-prefix/_next/static/chunks/04jv9e6~9vi.l.js","async":true}],["$","script","script-19",{"src":"/litellm-asset-prefix/_next/static/chunks/0d_sm.._5mw-p.js","async":true}],["$","script","script-20",{"src":"/litellm-asset-prefix/_next/static/chunks/0elr0ye86.44-.js","async":true}]],["$","$L6",null,{"children":["$","$7",null,{"name":"Next.MetadataOutlet","children":"$@8"}]}]]}],"isPartial":false,"staleTime":300,"varyParams":null,"buildId":"DSHomUr6Sq46Bm2WLdUas"}
0:{"rsc":["$","$1","c",{"children":[["$","$L2",null,{"Component":"$3","serverProvidedParams":{"searchParams":{},"params":{},"promises":["$@4","$@5"]}}],[["$","script","script-0",{"src":"/litellm-asset-prefix/_next/static/chunks/00g6xfr4yow7h.js","async":true}],["$","script","script-1",{"src":"/litellm-asset-prefix/_next/static/chunks/0-k_4_s7m108w.js","async":true}],["$","script","script-2",{"src":"/litellm-asset-prefix/_next/static/chunks/2kcxwg1mpncp6.js","async":true}],["$","script","script-3",{"src":"/litellm-asset-prefix/_next/static/chunks/3drq2_k-jeio2.js","async":true}],["$","script","script-4",{"src":"/litellm-asset-prefix/_next/static/chunks/395_vbpmrlvpu.js","async":true}],["$","script","script-5",{"src":"/litellm-asset-prefix/_next/static/chunks/2hu1vyy-5pv13.js","async":true}],["$","script","script-6",{"src":"/litellm-asset-prefix/_next/static/chunks/2l25bmiiw9ixp.js","async":true}],["$","script","script-7",{"src":"/litellm-asset-prefix/_next/static/chunks/1uz3jt-tj9lkf.js","async":true}],["$","script","script-8",{"src":"/litellm-asset-prefix/_next/static/chunks/199uwr871eene.js","async":true}],["$","script","script-9",{"src":"/litellm-asset-prefix/_next/static/chunks/2s_ce-opzrkzr.js","async":true}],["$","script","script-10",{"src":"/litellm-asset-prefix/_next/static/chunks/2ptdxz8qnchh_.js","async":true}],["$","script","script-11",{"src":"/litellm-asset-prefix/_next/static/chunks/17nqbxvhztf3k.js","async":true}],["$","script","script-12",{"src":"/litellm-asset-prefix/_next/static/chunks/2c90xukbd3il6.js","async":true}],["$","script","script-13",{"src":"/litellm-asset-prefix/_next/static/chunks/0kap_rdm2-lem.js","async":true}],["$","script","script-14",{"src":"/litellm-asset-prefix/_next/static/chunks/09l_m9l1emin2.js","async":true}],["$","script","script-15",{"src":"/litellm-asset-prefix/_next/static/chunks/1cea03gg5a_c7.js","async":true}],["$","script","script-16",{"src":"/litellm-asset-prefix/_next/static/chunks/323l6h8s7ahat.js","async":true}],["$","script","script-17",{"src":"/litellm-asset-prefix/_next/static/chunks/112n0hv3cc2rg.js","async":true}],["$","script","script-18",{"src":"/litellm-asset-prefix/_next/static/chunks/3y674jhwchpcq.js","async":true}],["$","script","script-19",{"src":"/litellm-asset-prefix/_next/static/chunks/105643dvf00hu.js","async":true}],["$","script","script-20",{"src":"/litellm-asset-prefix/_next/static/chunks/12wsfsljxg4xv.js","async":true}],["$","script","script-21",{"src":"/litellm-asset-prefix/_next/static/chunks/22iools_e0k44.js","async":true}],["$","script","script-22",{"src":"/litellm-asset-prefix/_next/static/chunks/3f9uewf5w-e-p.js","async":true}],["$","script","script-23",{"src":"/litellm-asset-prefix/_next/static/chunks/23-g73xaw3kap.js","async":true}]],["$","$L6",null,{"children":["$","$7",null,{"name":"Next.MetadataOutlet","children":"$@8"}]}]]}],"isPartial":false,"staleTime":300,"varyParams":null,"buildId":"0ljiPmkOdq7_yE4sZoXlJ"}
4:{}
5:"$0:rsc:props:children:0:props:serverProvidedParams:params"
8:null

10
litellm/proxy/_experimental/out/__next.!KGRhc2hib2FyZCk.txt
View file

@ -1,7 +1,7 @@
1:"$Sreact.fragment"
2:I[92825,["/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js"],"ClientSegmentRoot"]
3:I[216370,["/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js","/litellm-asset-prefix/_next/static/chunks/0mqbd99.ej13v.js","/litellm-asset-prefix/_next/static/chunks/16ufy1iyybswo.js","/litellm-asset-prefix/_next/static/chunks/0ayum-x.hkww~.js","/litellm-asset-prefix/_next/static/chunks/15jl-1gcakfwa.js","/litellm-asset-prefix/_next/static/chunks/003_1s9xbht43.js","/litellm-asset-prefix/_next/static/chunks/0u.r3vzo30ofk.js","/litellm-asset-prefix/_next/static/chunks/0g4qcx-c9gsxn.js","/litellm-asset-prefix/_next/static/chunks/14a-un1blorp~.js","/litellm-asset-prefix/_next/static/chunks/02_q4881cz6h~.js","/litellm-asset-prefix/_next/static/chunks/0lg.6rbfsd-l9.js","/litellm-asset-prefix/_next/static/chunks/0wdlbe750tuzr.js","/litellm-asset-prefix/_next/static/chunks/043q3g5-5-aju.js","/litellm-asset-prefix/_next/static/chunks/0nht59ws0elww.js","/litellm-asset-prefix/_next/static/chunks/03zkt5iyjiqcz.js","/litellm-asset-prefix/_next/static/chunks/0l02mpo6za6ie.js"],"default"]
4:I[339756,["/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js"],"default"]
5:I[837457,["/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js"],"default"]
0:{"rsc":["$","$1","c",{"children":[[["$","script","script-0",{"src":"/litellm-asset-prefix/_next/static/chunks/0mqbd99.ej13v.js","async":true}],["$","script","script-1",{"src":"/litellm-asset-prefix/_next/static/chunks/16ufy1iyybswo.js","async":true}],["$","script","script-2",{"src":"/litellm-asset-prefix/_next/static/chunks/0ayum-x.hkww~.js","async":true}],["$","script","script-3",{"src":"/litellm-asset-prefix/_next/static/chunks/15jl-1gcakfwa.js","async":true}],["$","script","script-4",{"src":"/litellm-asset-prefix/_next/static/chunks/003_1s9xbht43.js","async":true}],["$","script","script-5",{"src":"/litellm-asset-prefix/_next/static/chunks/0u.r3vzo30ofk.js","async":true}],["$","script","script-6",{"src":"/litellm-asset-prefix/_next/static/chunks/0g4qcx-c9gsxn.js","async":true}],["$","script","script-7",{"src":"/litellm-asset-prefix/_next/static/chunks/14a-un1blorp~.js","async":true}],["$","script","script-8",{"src":"/litellm-asset-prefix/_next/static/chunks/02_q4881cz6h~.js","async":true}],["$","script","script-9",{"src":"/litellm-asset-prefix/_next/static/chunks/0lg.6rbfsd-l9.js","async":true}],["$","script","script-10",{"src":"/litellm-asset-prefix/_next/static/chunks/0wdlbe750tuzr.js","async":true}],["$","script","script-11",{"src":"/litellm-asset-prefix/_next/static/chunks/043q3g5-5-aju.js","async":true}],["$","script","script-12",{"src":"/litellm-asset-prefix/_next/static/chunks/0nht59ws0elww.js","async":true}],["$","script","script-13",{"src":"/litellm-asset-prefix/_next/static/chunks/03zkt5iyjiqcz.js","async":true}],["$","script","script-14",{"src":"/litellm-asset-prefix/_next/static/chunks/0l02mpo6za6ie.js","async":true}]],["$","$L2",null,{"Component":"$3","slots":{"children":["$","$L4",null,{"parallelRouterKey":"children","template":["$","$L5",null,{}],"notFound":[[["$","title",null,{"children":"404: This page could not be found."}],["$","div",null,{"style":{"fontFamily":"system-ui,\"Segoe UI\",Roboto,Helvetica,Arial,sans-serif,\"Apple Color Emoji\",\"Segoe UI Emoji\"","height":"100vh","textAlign":"center","display":"flex","flexDirection":"column","alignItems":"center","justifyContent":"center"},"children":["$","div",null,{"children":[["$","style",null,{"dangerouslySetInnerHTML":{"__html":"body{color:#000;background:#fff;margin:0}.next-error-h1{border-right:1px solid rgba(0,0,0,.3)}@media (prefers-color-scheme:dark){body{color:#fff;background:#000}.next-error-h1{border-right:1px solid rgba(255,255,255,.3)}}"}}],["$","h1",null,{"className":"next-error-h1","style":{"display":"inline-block","margin":"0 20px 0 0","padding":"0 23px 0 0","fontSize":24,"fontWeight":500,"verticalAlign":"top","lineHeight":"49px"},"children":404}],["$","div",null,{"style":{"display":"inline-block"},"children":["$","h2",null,{"style":{"fontSize":14,"fontWeight":400,"lineHeight":"49px","margin":0},"children":"This page could not be found."}]}]]}]}]],[]]}]},"serverProvidedParams":{"params":{},"promises":["$@6"]}}]]}],"isPartial":false,"staleTime":300,"varyParams":null,"buildId":"DSHomUr6Sq46Bm2WLdUas"}
2:I[92825,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"ClientSegmentRoot"]
3:I[216370,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js","/litellm-asset-prefix/_next/static/chunks/1ioy8obpggx93.js","/litellm-asset-prefix/_next/static/chunks/3_3dj4vdy-3xy.js","/litellm-asset-prefix/_next/static/chunks/1zr7rrk4wkmju.js","/litellm-asset-prefix/_next/static/chunks/2cngn5bal3278.js","/litellm-asset-prefix/_next/static/chunks/2up3bks93iqds.js","/litellm-asset-prefix/_next/static/chunks/1l2mgm5v3tjci.js","/litellm-asset-prefix/_next/static/chunks/0zduf1gntl_f8.js","/litellm-asset-prefix/_next/static/chunks/0g_w4tf2inv3i.js","/litellm-asset-prefix/_next/static/chunks/1a0bgy7kzrj91.js","/litellm-asset-prefix/_next/static/chunks/0dbvgsc7ha049.js","/litellm-asset-prefix/_next/static/chunks/17-6zku8f68gf.js","/litellm-asset-prefix/_next/static/chunks/0f5fel02jwglw.js","/litellm-asset-prefix/_next/static/chunks/1vquuz09jxl5_.js","/litellm-asset-prefix/_next/static/chunks/1iakmimqrlpn0.js","/litellm-asset-prefix/_next/static/chunks/2n26sdz53rm0a.js","/litellm-asset-prefix/_next/static/chunks/11khk745tfruy.js"],"default"]
4:I[339756,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"default"]
5:I[837457,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"default"]
0:{"rsc":["$","$1","c",{"children":[[["$","script","script-0",{"src":"/litellm-asset-prefix/_next/static/chunks/1ioy8obpggx93.js","async":true}],["$","script","script-1",{"src":"/litellm-asset-prefix/_next/static/chunks/3_3dj4vdy-3xy.js","async":true}],["$","script","script-2",{"src":"/litellm-asset-prefix/_next/static/chunks/1zr7rrk4wkmju.js","async":true}],["$","script","script-3",{"src":"/litellm-asset-prefix/_next/static/chunks/2cngn5bal3278.js","async":true}],["$","script","script-4",{"src":"/litellm-asset-prefix/_next/static/chunks/2up3bks93iqds.js","async":true}],["$","script","script-5",{"src":"/litellm-asset-prefix/_next/static/chunks/1l2mgm5v3tjci.js","async":true}],["$","script","script-6",{"src":"/litellm-asset-prefix/_next/static/chunks/0zduf1gntl_f8.js","async":true}],["$","script","script-7",{"src":"/litellm-asset-prefix/_next/static/chunks/0g_w4tf2inv3i.js","async":true}],["$","script","script-8",{"src":"/litellm-asset-prefix/_next/static/chunks/1a0bgy7kzrj91.js","async":true}],["$","script","script-9",{"src":"/litellm-asset-prefix/_next/static/chunks/0dbvgsc7ha049.js","async":true}],["$","script","script-10",{"src":"/litellm-asset-prefix/_next/static/chunks/17-6zku8f68gf.js","async":true}],["$","script","script-11",{"src":"/litellm-asset-prefix/_next/static/chunks/0f5fel02jwglw.js","async":true}],["$","script","script-12",{"src":"/litellm-asset-prefix/_next/static/chunks/1vquuz09jxl5_.js","async":true}],["$","script","script-13",{"src":"/litellm-asset-prefix/_next/static/chunks/1iakmimqrlpn0.js","async":true}],["$","script","script-14",{"src":"/litellm-asset-prefix/_next/static/chunks/2n26sdz53rm0a.js","async":true}],["$","script","script-15",{"src":"/litellm-asset-prefix/_next/static/chunks/11khk745tfruy.js","async":true}]],["$","$L2",null,{"Component":"$3","slots":{"children":["$","$L4",null,{"parallelRouterKey":"children","template":["$","$L5",null,{}],"notFound":[[["$","title",null,{"children":"404: This page could not be found."}],["$","div",null,{"style":{"fontFamily":"system-ui,\"Segoe UI\",Roboto,Helvetica,Arial,sans-serif,\"Apple Color Emoji\",\"Segoe UI Emoji\"","height":"100vh","textAlign":"center","display":"flex","flexDirection":"column","alignItems":"center","justifyContent":"center"},"children":["$","div",null,{"children":[["$","style",null,{"dangerouslySetInnerHTML":{"__html":"body{color:#000;background:#fff;margin:0}.next-error-h1{border-right:1px solid rgba(0,0,0,.3)}@media (prefers-color-scheme:dark){body{color:#fff;background:#000}.next-error-h1{border-right:1px solid rgba(255,255,255,.3)}}"}}],["$","h1",null,{"className":"next-error-h1","style":{"display":"inline-block","margin":"0 20px 0 0","padding":"0 23px 0 0","fontSize":24,"fontWeight":500,"verticalAlign":"top","lineHeight":"49px"},"children":404}],["$","div",null,{"style":{"display":"inline-block"},"children":["$","h2",null,{"style":{"fontSize":14,"fontWeight":400,"lineHeight":"49px","margin":0},"children":"This page could not be found."}]}]]}]}]],[]]}]},"serverProvidedParams":{"params":{},"promises":["$@6"]}}]]}],"isPartial":false,"staleTime":300,"varyParams":null,"buildId":"0ljiPmkOdq7_yE4sZoXlJ"}
6:"$0:rsc:props:children:1:props:serverProvidedParams:params"

61
litellm/proxy/_experimental/out/__next._full.txt
View file

File diff suppressed because one or more lines are too long

8
litellm/proxy/_experimental/out/__next._head.txt
View file

@ -1,6 +1,6 @@
1:"$Sreact.fragment"
2:I[897367,["/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js"],"ViewportBoundary"]
3:I[897367,["/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js"],"MetadataBoundary"]
2:I[897367,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"ViewportBoundary"]
3:I[897367,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"MetadataBoundary"]
4:"$Sreact.suspense"
5:I[27201,["/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js"],"IconMark"]
0:{"rsc":["$","$1","h",{"children":[null,["$","$L2",null,{"children":[["$","meta","0",{"charSet":"utf-8"}],["$","meta","1",{"name":"viewport","content":"width=device-width, initial-scale=1"}]]}],["$","div",null,{"hidden":true,"children":["$","$L3",null,{"children":["$","$4",null,{"name":"Next.Metadata","children":[["$","title","0",{"children":"LiteLLM Dashboard"}],["$","meta","1",{"name":"description","content":"LiteLLM Proxy Admin UI"}],["$","link","2",{"rel":"icon","href":"/favicon.ico?favicon.0~dgapwhi~75y.ico","sizes":"48x48","type":"image/x-icon"}],["$","link","3",{"rel":"icon","href":"/get_favicon"}],["$","$L5","4",{}]]}]}]}],["$","meta",null,{"name":"next-size-adjust","content":""}]]}],"isPartial":false,"staleTime":300,"varyParams":null,"buildId":"DSHomUr6Sq46Bm2WLdUas"}
5:I[27201,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"IconMark"]
0:{"rsc":["$","$1","h",{"children":[null,["$","$L2",null,{"children":[["$","meta","0",{"charSet":"utf-8"}],["$","meta","1",{"name":"viewport","content":"width=device-width, initial-scale=1"}]]}],["$","div",null,{"hidden":true,"children":["$","$L3",null,{"children":["$","$4",null,{"name":"Next.Metadata","children":[["$","title","0",{"children":"LiteLLM Dashboard"}],["$","meta","1",{"name":"description","content":"LiteLLM Proxy Admin UI"}],["$","link","2",{"rel":"icon","href":"/favicon.ico?favicon.3arlap5n8tyzg.ico","sizes":"48x48","type":"image/x-icon"}],["$","link","3",{"rel":"icon","href":"/get_favicon"}],["$","$L5","4",{}]]}]}]}],["$","meta",null,{"name":"next-size-adjust","content":""}]]}],"isPartial":false,"staleTime":300,"varyParams":null,"buildId":"0ljiPmkOdq7_yE4sZoXlJ"}

16
litellm/proxy/_experimental/out/__next._index.txt
View file

@ -1,9 +1,9 @@
1:"$Sreact.fragment"
2:I[867271,["/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js"],"default"]
3:I[71195,["/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js"],"default"]
4:I[557951,["/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js"],"AuthProvider"]
5:I[339756,["/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js"],"default"]
6:I[837457,["/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js"],"default"]
:HL["/litellm-asset-prefix/_next/static/chunks/05qmwjqau64bz.css","style"]
:HL["/litellm-asset-prefix/_next/static/chunks/0aapv6n5bztwf.css","style"]
0:{"rsc":["$","$1","c",{"children":[[["$","link","0",{"rel":"stylesheet","href":"/litellm-asset-prefix/_next/static/chunks/05qmwjqau64bz.css","precedence":"next"}],["$","link","1",{"rel":"stylesheet","href":"/litellm-asset-prefix/_next/static/chunks/0aapv6n5bztwf.css","precedence":"next"}],["$","script","script-0",{"src":"/litellm-asset-prefix/_next/static/chunks/09qhs_ev_bxr8.js","async":true}],["$","script","script-1",{"src":"/litellm-asset-prefix/_next/static/chunks/0l6q6-77u4dy~.js","async":true}],["$","script","script-2",{"src":"/litellm-asset-prefix/_next/static/chunks/0hpxif-db_y5-.js","async":true}]],["$","html",null,{"lang":"en","children":["$","body",null,{"className":"inter_5972bc34-module__OU16Qa__className","children":["$","$L2",null,{"children":["$","$L3",null,{"children":["$","$L4",null,{"children":["$","$L5",null,{"parallelRouterKey":"children","template":["$","$L6",null,{}],"notFound":[[["$","title",null,{"children":"404: This page could not be found."}],["$","div",null,{"style":{"fontFamily":"system-ui,\"Segoe UI\",Roboto,Helvetica,Arial,sans-serif,\"Apple Color Emoji\",\"Segoe UI Emoji\"","height":"100vh","textAlign":"center","display":"flex","flexDirection":"column","alignItems":"center","justifyContent":"center"},"children":["$","div",null,{"children":[["$","style",null,{"dangerouslySetInnerHTML":{"__html":"body{color:#000;background:#fff;margin:0}.next-error-h1{border-right:1px solid rgba(0,0,0,.3)}@media (prefers-color-scheme:dark){body{color:#fff;background:#000}.next-error-h1{border-right:1px solid rgba(255,255,255,.3)}}"}}],["$","h1",null,{"className":"next-error-h1","style":{"display":"inline-block","margin":"0 20px 0 0","padding":"0 23px 0 0","fontSize":24,"fontWeight":500,"verticalAlign":"top","lineHeight":"49px"},"children":404}],["$","div",null,{"style":{"display":"inline-block"},"children":["$","h2",null,{"style":{"fontSize":14,"fontWeight":400,"lineHeight":"49px","margin":0},"children":"This page could not be found."}]}]]}]}]],[]]}]}]}]}]}]}]]}],"isPartial":false,"staleTime":300,"varyParams":null,"buildId":"DSHomUr6Sq46Bm2WLdUas"}
2:I[867271,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"default"]
3:I[71195,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"default"]
4:I[557951,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"AuthProvider"]
5:I[339756,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"default"]
6:I[837457,["/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js"],"default"]
:HL["/litellm-asset-prefix/_next/static/chunks/1kid9zr1--h6y.css","style"]
:HL["/litellm-asset-prefix/_next/static/chunks/23vtcpdpp2h9h.css","style"]
0:{"rsc":["$","$1","c",{"children":[[["$","link","0",{"rel":"stylesheet","href":"/litellm-asset-prefix/_next/static/chunks/1kid9zr1--h6y.css","precedence":"next"}],["$","link","1",{"rel":"stylesheet","href":"/litellm-asset-prefix/_next/static/chunks/23vtcpdpp2h9h.css","precedence":"next"}],["$","script","script-0",{"src":"/litellm-asset-prefix/_next/static/chunks/0vggytdohwe7o.js","async":true}],["$","script","script-1",{"src":"/litellm-asset-prefix/_next/static/chunks/0map77ee0fk0e.js","async":true}],["$","script","script-2",{"src":"/litellm-asset-prefix/_next/static/chunks/1jfookxfajkeo.js","async":true}]],["$","html",null,{"lang":"en","children":["$","body",null,{"className":"inter_5972bc34-module__OU16Qa__className","children":["$","$L2",null,{"children":["$","$L3",null,{"children":["$","$L4",null,{"children":["$","$L5",null,{"parallelRouterKey":"children","template":["$","$L6",null,{}],"notFound":[[["$","title",null,{"children":"404: This page could not be found."}],["$","div",null,{"style":{"fontFamily":"system-ui,\"Segoe UI\",Roboto,Helvetica,Arial,sans-serif,\"Apple Color Emoji\",\"Segoe UI Emoji\"","height":"100vh","textAlign":"center","display":"flex","flexDirection":"column","alignItems":"center","justifyContent":"center"},"children":["$","div",null,{"children":[["$","style",null,{"dangerouslySetInnerHTML":{"__html":"body{color:#000;background:#fff;margin:0}.next-error-h1{border-right:1px solid rgba(0,0,0,.3)}@media (prefers-color-scheme:dark){body{color:#fff;background:#000}.next-error-h1{border-right:1px solid rgba(255,255,255,.3)}}"}}],["$","h1",null,{"className":"next-error-h1","style":{"display":"inline-block","margin":"0 20px 0 0","padding":"0 23px 0 0","fontSize":24,"fontWeight":500,"verticalAlign":"top","lineHeight":"49px"},"children":404}],["$","div",null,{"style":{"display":"inline-block"},"children":["$","h2",null,{"style":{"fontSize":14,"fontWeight":400,"lineHeight":"49px","margin":0},"children":"This page could not be found."}]}]]}]}]],[]]}]}]}]}]}]}]]}],"isPartial":false,"staleTime":300,"varyParams":null,"buildId":"0ljiPmkOdq7_yE4sZoXlJ"}

8
litellm/proxy/_experimental/out/__next._tree.txt
View file

@ -1,4 +1,4 @@
:HL["/litellm-asset-prefix/_next/static/chunks/05qmwjqau64bz.css","style"]
:HL["/litellm-asset-prefix/_next/static/chunks/0aapv6n5bztwf.css","style"]
:HL["/litellm-asset-prefix/_next/static/media/83afe278b6a6bb3c-s.p.0q-301v4kxxnr.woff2","font",{"crossOrigin":"","type":"font/woff2"}]
0:{"tree":{"name":"","param":null,"prefetchHints":16,"slots":{"children":{"name":"(dashboard)","param":null,"prefetchHints":0,"slots":{"children":{"name":"__PAGE__","param":null,"prefetchHints":0,"slots":null}}}}},"staleTime":300,"buildId":"DSHomUr6Sq46Bm2WLdUas"}
:HL["/litellm-asset-prefix/_next/static/chunks/1kid9zr1--h6y.css","style"]
:HL["/litellm-asset-prefix/_next/static/chunks/23vtcpdpp2h9h.css","style"]
:HL["/litellm-asset-prefix/_next/static/media/83afe278b6a6bb3c-s.p.2bn3s6zvc0dyp.woff2","font",{"crossOrigin":"","type":"font/woff2"}]
0:{"tree":{"name":"","param":null,"prefetchHints":16,"slots":{"children":{"name":"(dashboard)","param":null,"prefetchHints":0,"slots":{"children":{"name":"__PAGE__","param":null,"prefetchHints":0,"slots":null}}}}},"staleTime":300,"buildId":"0ljiPmkOdq7_yE4sZoXlJ"}

0
litellm/proxy/_experimental/out/_next/static/DSHomUr6Sq46Bm2WLdUas/_buildManifest.js → litellm/proxy/_experimental/out/_next/static/0ljiPmkOdq7_yE4sZoXlJ/_buildManifest.js
View file

0
litellm/proxy/_experimental/out/_next/static/DSHomUr6Sq46Bm2WLdUas/_clientMiddlewareManifest.js → litellm/proxy/_experimental/out/_next/static/0ljiPmkOdq7_yE4sZoXlJ/_clientMiddlewareManifest.js
View file

0
litellm/proxy/_experimental/out/_next/static/DSHomUr6Sq46Bm2WLdUas/_ssgManifest.js → litellm/proxy/_experimental/out/_next/static/0ljiPmkOdq7_yE4sZoXlJ/_ssgManifest.js
View file

1
litellm/proxy/_experimental/out/_next/static/chunks/0-_m4km7b1~oe.js
View file

File diff suppressed because one or more lines are too long

8
litellm/proxy/_experimental/out/_next/static/chunks/0-ahu72ndvhwn.js
View file

File diff suppressed because one or more lines are too long

7
litellm/proxy/_experimental/out/_next/static/chunks/0-k_4_s7m108w.js Normal file
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/0.3q2b74j~ty5.js
View file

File diff suppressed because one or more lines are too long

2
litellm/proxy/_experimental/out/_next/static/chunks/0.mwuwep0859t.js
View file

File diff suppressed because one or more lines are too long

10
litellm/proxy/_experimental/out/_next/static/chunks/0.p~s6ih~c~xe.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/00-dyuivh_bf-.js Normal file
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/003_1s9xbht43.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/003w1n3_ylv_2.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/007c8g8hmd9qz.js Normal file
View file

File diff suppressed because one or more lines are too long

8
litellm/proxy/_experimental/out/_next/static/chunks/00ccjtnk99zr7.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/00g6xfr4yow7h.js Normal file
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/00qiry~y.broe.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/00tczcrtv5upo.js
View file

File diff suppressed because one or more lines are too long

8
litellm/proxy/_experimental/out/_next/static/chunks/00zxtugv201bq.js
View file

File diff suppressed because one or more lines are too long

16
litellm/proxy/_experimental/out/_next/static/chunks/0175usbyz91lt.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/017kxo-8o84bv.js Normal file
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/01hy_w_4bnb34.js
View file

File diff suppressed because one or more lines are too long

8
litellm/proxy/_experimental/out/_next/static/chunks/01ozl298h03bw.js
View file

File diff suppressed because one or more lines are too long

4
litellm/proxy/_experimental/out/_next/static/chunks/01y._o853f7le.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/01yk5y7rumzgt.js
View file

File diff suppressed because one or more lines are too long

4
litellm/proxy/_experimental/out/_next/static/chunks/022.sz94ycw4x.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/023jsye4cz4a7.js
View file

File diff suppressed because one or more lines are too long

2
litellm/proxy/_experimental/out/_next/static/chunks/026n9mracjd5k.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/027d2u2cl335o.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/02_q4881cz6h~.js
View file

File diff suppressed because one or more lines are too long

2
litellm/proxy/_experimental/out/_next/static/chunks/02nioff5-e.ez.js
View file

File diff suppressed because one or more lines are too long

17
litellm/proxy/_experimental/out/_next/static/chunks/02ucg1k1-nq5m.js Normal file
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/02wxbd2ona7u_.js
View file

File diff suppressed because one or more lines are too long

68
litellm/proxy/_experimental/out/_next/static/chunks/030xj-a9q0ur8.js Normal file
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/0337vg5sc7rt~.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/0369tkoo6z4yx.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/036yal3~xlgjh.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/03oh9wvqpsr-g.js
View file

File diff suppressed because one or more lines are too long

31
litellm/proxy/_experimental/out/_next/static/chunks/03sdszpwi459j.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/03zkt5iyjiqcz.js
View file

File diff suppressed because one or more lines are too long

56
litellm/proxy/_experimental/out/_next/static/chunks/04.hopkzyt7jd.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/046-gw19n7owc.js
View file

File diff suppressed because one or more lines are too long

8
litellm/proxy/_experimental/out/_next/static/chunks/04_xp3aju8b3x.js
View file

File diff suppressed because one or more lines are too long

2
litellm/proxy/_experimental/out/_next/static/chunks/04jv9e6~9vi.l.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/04m0obyskflau.js Normal file
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/04rayq7y4j4oi.js
View file

File diff suppressed because one or more lines are too long

2
litellm/proxy/_experimental/out/_next/static/chunks/12eumif3gapzm.js → litellm/proxy/_experimental/out/_next/static/chunks/04y2hqzy08peg.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/05id71gg6oywc.js Normal file
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/05wd9su61xvp4.js
View file

File diff suppressed because one or more lines are too long

4
litellm/proxy/_experimental/out/_next/static/chunks/05z02g9s~8km0.js
View file

File diff suppressed because one or more lines are too long

216
litellm/proxy/_experimental/out/_next/static/chunks/060kl3yana4g8.js
View file

File diff suppressed because one or more lines are too long

35
litellm/proxy/_experimental/out/_next/static/chunks/068p6o.s_qzmk.js
View file

@ -1,35 +0,0 @@
(globalThis.TURBOPACK||(globalThis.TURBOPACK=[])).push(["object"==typeof document?document.currentScript:void 0,95779,e=>{"use strict";var t=e.i(480731);let r=[t.BaseColors.Blue,t.BaseColors.Cyan,t.BaseColors.Sky,t.BaseColors.Indigo,t.BaseColors.Violet,t.BaseColors.Purple,t.BaseColors.Fuchsia,t.BaseColors.Slate,t.BaseColors.Gray,t.BaseColors.Zinc,t.BaseColors.Neutral,t.BaseColors.Stone,t.BaseColors.Red,t.BaseColors.Orange,t.BaseColors.Amber,t.BaseColors.Yellow,t.BaseColors.Lime,t.BaseColors.Green,t.BaseColors.Emerald,t.BaseColors.Teal,t.BaseColors.Pink,t.BaseColors.Rose];e.s(["colorPalette",0,{canvasBackground:50,lightBackground:100,background:500,darkBackground:600,darkestBackground:800,lightBorder:200,border:500,darkBorder:700,lightRing:200,ring:300,iconRing:500,lightText:400,text:500,iconText:600,darkText:700,darkestText:900,icon:500},"themeColorRange",0,r])},629569,e=>{"use strict";var t=e.i(290571),r=e.i(95779),o=e.i(444755),s=e.i(673706),l=e.i(271645);let a=l.default.forwardRef((e,a)=>{let{color:i,children:n,className:d}=e,c=(0,t.__rest)(e,["color","children","className"]);return l.default.createElement("p",Object.assign({ref:a,className:(0,o.tremorTwMerge)("font-medium text-tremor-title",i?(0,s.getColorClassNames)(i,r.colorPalette.darkText).textColor:"text-tremor-content-strong dark:text-dark-tremor-content-strong",d)},c),n)});a.displayName="Title",e.s(["Title",0,a],629569)},166406,e=>{"use strict";var t=e.i(190144);e.s(["CopyOutlined",()=>t.default])},411929,e=>{"use strict";var t=e.i(843476),r=e.i(271645),o=e.i(464571),s=e.i(166406),l=e.i(629569),a=e.i(602869),i=e.i(727749);let n=({accessToken:e})=>{let[n,d]=(0,r.useState)(`{
"model": "openai/gpt-4o",
"messages": [
{
"role": "system",
"content": "You are a helpful assistant."
},
{
"role": "user",
"content": "Explain quantum computing in simple terms"
}
],
"temperature": 0.7,
"max_tokens": 500,
"stream": true
}`),[c,u]=(0,r.useState)(""),[p,x]=(0,r.useState)(!1),m=async()=>{x(!0);try{let s;try{s=JSON.parse(n)}catch(e){i.default.fromBackend("Invalid JSON in request body"),x(!1);return}let l={call_type:"completion",request_body:s};if(!e){i.default.fromBackend("No access token found"),x(!1);return}let d=await (0,a.transformRequestCall)(e,l);if(d.raw_request_api_base&&d.raw_request_body){var t,r,o;let e,s,l=(t=d.raw_request_api_base,r=d.raw_request_body,o=d.raw_request_headers||{},e=JSON.stringify(r,null,2).split("\n").map(e=>` ${e}`).join("\n"),s=Object.entries(o).map(([e,t])=>`-H '${e}: ${t}'`).join(" \\\n "),`curl -X POST \\
${t} \\
${s?`${s} \\
`:""}-H 'Content-Type: application/json' \\
-d '{
${e}
}'`);u(l),i.default.success("Request transformed successfully")}else{let e="string"==typeof d?d:JSON.stringify(d);u(e),i.default.info("Transformed request received in unexpected format")}}catch(e){console.error("Error transforming request:",e),i.default.fromBackend("Failed to transform request")}finally{x(!1)}};return(0,t.jsxs)("div",{className:"w-full m-2",style:{overflow:"hidden"},children:[(0,t.jsx)(l.Title,{children:"Playground"}),(0,t.jsx)("p",{className:"text-sm text-gray-500",children:"See how LiteLLM transforms your request for the specified provider."}),(0,t.jsxs)("div",{style:{display:"flex",gap:"16px",width:"100%",minWidth:0,overflow:"hidden"},className:"mt-4",children:[(0,t.jsxs)("div",{style:{flex:"1 1 50%",display:"flex",flexDirection:"column",border:"1px solid #e8e8e8",borderRadius:"8px",padding:"24px",overflow:"hidden",maxHeight:"600px",minWidth:0},children:[(0,t.jsxs)("div",{style:{marginBottom:"24px"},children:[(0,t.jsx)("h2",{style:{fontSize:"24px",fontWeight:"bold",margin:"0 0 4px 0"},children:"Original Request"}),(0,t.jsx)("p",{style:{color:"#666",margin:0},children:"The request you would send to LiteLLM /chat/completions endpoint."})]}),(0,t.jsx)("textarea",{style:{flex:"1 1 auto",width:"100%",minHeight:"240px",padding:"16px",border:"1px solid #e8e8e8",borderRadius:"6px",fontFamily:"monospace",fontSize:"14px",resize:"none",marginBottom:"24px",overflow:"auto"},value:n,onChange:e=>d(e.target.value),onKeyDown:e=>{(e.metaKey||e.ctrlKey)&&"Enter"===e.key&&(e.preventDefault(),m())},placeholder:"Press Cmd/Ctrl + Enter to transform"}),(0,t.jsx)("div",{style:{display:"flex",justifyContent:"flex-end",marginTop:"auto"},children:(0,t.jsxs)(o.Button,{type:"primary",style:{backgroundColor:"#000",display:"flex",alignItems:"center",gap:"8px"},onClick:m,loading:p,children:[(0,t.jsx)("span",{children:"Transform"}),(0,t.jsx)("span",{children:"→"})]})})]}),(0,t.jsxs)("div",{style:{flex:"1 1 50%",display:"flex",flexDirection:"column",border:"1px solid #e8e8e8",borderRadius:"8px",padding:"24px",overflow:"hidden",maxHeight:"800px",minWidth:0},children:[(0,t.jsxs)("div",{style:{marginBottom:"24px"},children:[(0,t.jsx)("h2",{style:{fontSize:"24px",fontWeight:"bold",margin:"0 0 4px 0"},children:"Transformed Request"}),(0,t.jsx)("p",{style:{color:"#666",margin:0},children:"How LiteLLM transforms your request for the specified provider."}),(0,t.jsx)("br",{}),(0,t.jsx)("p",{style:{color:"#666",margin:0},className:"text-xs",children:"Note: Sensitive headers are not shown."})]}),(0,t.jsxs)("div",{style:{position:"relative",backgroundColor:"#f5f5f5",borderRadius:"6px",flex:"1 1 auto",display:"flex",flexDirection:"column",overflow:"hidden"},children:[(0,t.jsx)("pre",{style:{padding:"16px",fontFamily:"monospace",fontSize:"14px",margin:0,overflow:"auto",flex:"1 1 auto"},children:c||`curl -X POST \\
https://api.openai.com/v1/chat/completions \\
-H 'Authorization: Bearer sk-xxx' \\
-H 'Content-Type: application/json' \\
-d '{
"model": "gpt-4",
"messages": [
{
"role": "system",
"content": "You are a helpful assistant."
}
],
"temperature": 0.7
}'`}),(0,t.jsx)(o.Button,{type:"text",icon:(0,t.jsx)(s.CopyOutlined,{}),style:{position:"absolute",right:"8px",top:"8px"},size:"small",onClick:()=>{navigator.clipboard.writeText(c||""),i.default.success("Copied to clipboard")}})]})]})]}),(0,t.jsx)("div",{className:"mt-4 text-right w-full",children:(0,t.jsxs)("p",{className:"text-sm text-gray-500",children:["Found an error? File an issue"," ",(0,t.jsx)("a",{href:"https://github.com/BerriAI/litellm/issues",target:"_blank",rel:"noopener noreferrer",children:"here"}),"."]})})]})};var d=e.i(135214);e.s(["default",0,function(){let{accessToken:e}=(0,d.default)();return(0,t.jsx)(n,{accessToken:e})}],411929)}]);

1
litellm/proxy/_experimental/out/_next/static/chunks/069dx5~5osue0.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/06_bx9tq0eg6t.js
View file

File diff suppressed because one or more lines are too long

17
litellm/proxy/_experimental/out/_next/static/chunks/06d3gjz2_.wju.js
View file

File diff suppressed because one or more lines are too long

420
litellm/proxy/_experimental/out/_next/static/chunks/06f~oqn5wl_jt.js
View file

File diff suppressed because one or more lines are too long

1
litellm/proxy/_experimental/out/_next/static/chunks/06rg~x2ihanj..js
View file

File diff suppressed because one or more lines are too long

13
litellm/proxy/_experimental/out/_next/static/chunks/06v.xgo7n3be4.js
View file

File diff suppressed because one or more lines are too long

10
litellm/proxy/_experimental/out/_next/static/chunks/06xk.10xipp8w.js
View file

File diff suppressed because one or more lines are too long

Some files were not shown because too many files have changed in this diff Show more