Commit graph

34557 commits

Author SHA1 Message Date
shin-bot-litellm
3ec91d4333
fix(ci): Fix ruff lint error - unused import in vertex_ai_ingestion (#21178)
Co-authored-by: shin-bot-litellm <shin-bot-litellm@users.noreply.github.com>
2026-02-13 18:14:23 -08:00
shin-bot-litellm
df54e1b452
fix(model_info): Add missing tpm/rpm for Gemini models (#21175)
Several Gemini models (TTS, native-audio, robotics, gemma) were missing
tpm/rpm values, causing test_get_model_info_gemini to fail.

Added conservative default values (tpm=250000, rpm=10) for preview models.
gemini-2.5-flash-preview-tts gets tpm=4000000, rpm=10.

Co-authored-by: OpenClaw <openclaw@users.noreply.github.com>
2026-02-13 18:07:01 -08:00
Alexsander Hamir
30b28da2b7
Add pyroscope for observability (#21167)
* Pyroscope: require PYROSCOPE_APP_NAME and PYROSCOPE_SERVER_ADDRESS, add UTF-8 locale hint

- No defaults for PYROSCOPE_APP_NAME or PYROSCOPE_SERVER_ADDRESS; fail at startup if unset when Pyroscope is enabled
- Set LANG/LC_ALL to C.UTF-8 when unset to reduce malformed_profile (invalid UTF-8) rejections
- Startup message suggests PYTHONUTF8=1 if server rejects profiles
- Simplify LITELLM_ENABLE_PYROSCOPE in config_settings; document Pyroscope env vars as required with no default
- Add pyroscope_profiling to sidebar (Alerting & Monitoring)
- pyproject.toml: pyroscope-io as required dep on non-Windows (marker), in proxy extra

* proxy: add PYROSCOPE_SAMPLE_RATE env, use verbose logging, fix int type

- Add optional PYROSCOPE_SAMPLE_RATE env (integer, no default)
- Pass sample_rate to pyroscope.configure() as int for pyroscope-io
- Replace print with verbose_proxy_logger (info/warning)
- Document PYROSCOPE_SAMPLE_RATE in config_settings.md

* Address Greptile PR feedback: Pyroscope optional, docs, tests, docstring

- pyproject.toml: mark pyroscope-io as optional=true (proxy extra only)
- Add docs/my-website/docs/proxy/pyroscope_profiling.md (fix broken sidebar link)
- Add tests/test_litellm/proxy/test_pyroscope.py for _init_pyroscope()
- proxy_server: fix _init_pyroscope docstring (required server/app name, sample rate as int)

* Update litellm/proxy/proxy_server.py

Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>

---------

Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
2026-02-13 17:32:29 -08:00
yuneng-jiang
1828e68515
Merge pull request #21166 from BerriAI/litellm_access_group_rename
[Refactor] Access Group model_ids to model_names for backwards Compatability
2026-02-13 17:22:17 -08:00
yuneng-jiang
3ebafc4a10
Merge pull request #21165 from BerriAI/litellm_access_group_ui
[Feature] UI - Access Groups: Table and Details Page
2026-02-13 17:17:18 -08:00
yuneng-jiang
41445ff8ad migration + build files 2026-02-13 17:15:39 -08:00
yuneng-jiang
38e2ecb52c bump: version 0.4.36 → 0.4.37 2026-02-13 17:15:05 -08:00
yuneng-jiang
60954d22e3 Merge remote-tracking branch 'origin' into litellm_access_group_rename 2026-02-13 17:04:59 -08:00
Ishaan Jaff
b48bec0e4b
fix: populate identity fields in proxy admin JWT early-return path (#21169)
* fix: populate identity fields in proxy admin JWT early-return path

When is_proxy_admin is True, the UserAPIKeyAuth early-return now includes
user_id, team_id, team_alias, team_metadata, org_id, and end_user_id
resolved from the JWT. Previously only user_role and parent_otel_span
were set, causing blank Team Name and Internal User in Request Logs UI.

* test: add unit tests for proxy admin JWT identity fields
2026-02-13 17:04:38 -08:00
yuneng-jiang
ea2e5ff0b9 allow editing of access group names 2026-02-13 17:03:41 -08:00
yuneng-jiang
7561c1a3c3 addressing comments 2026-02-13 16:54:34 -08:00
yuneng-jiang
72848f4c08 change to model name for backwards compat 2026-02-13 16:44:49 -08:00
The Mavik
ab4b6197ef
fix: add custom_body parameter to endpoint_func in create_pass_through_route (#20849)
* fix: add custom_body parameter to endpoint_func in create_pass_through_route

The bedrock_proxy_route calls `endpoint_func(custom_body=data)` to
pass a pre-parsed, SigV4-signed request body. However, the
`endpoint_func` closure created by `create_pass_through_route` does
not accept a `custom_body` keyword argument, causing:

    TypeError: endpoint_func() got an unexpected keyword argument 'custom_body'

Add `custom_body: Optional[dict] = None` to both `endpoint_func`
definitions (adapter-based and URL-based). In the URL-based path,
when `custom_body` is provided by the caller, use it instead of
re-parsing the body from the raw request.

Fixes #16999

* Add tests for custom_body handling in create_pass_through_route

Address reviewer feedback on PR #20849:

- Document why the adapter-based endpoint_func accepts custom_body
  for signature compatibility but does not forward it (the underlying
  chat_completion_pass_through_endpoint does not support it).
- Add test_create_pass_through_route_custom_body_url_target: verifies
  that when a caller (e.g. bedrock_proxy_route) supplies custom_body,
  it takes precedence over the body parsed from the raw request.
- Add test_create_pass_through_route_no_custom_body_falls_back:
  verifies that the default path (no custom_body) correctly uses the
  request-parsed body, preserving existing behavior.

Both tests are fully mocked following the project's CONTRIBUTING.md
guidelines and the patterns established in the existing test file.

Co-authored-by: Cursor <cursoragent@cursor.com>

---------

Co-authored-by: themavik <themavik@users.noreply.github.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-02-13 16:44:40 -08:00
yuneng-jiang
ca2ce4a946 adding tests 2026-02-13 16:30:24 -08:00
yuneng-jiang
2f3dc4860f new badge changes 2026-02-13 16:22:07 -08:00
yuneng-jiang
5dbcca8d43 Access groups UI 2026-02-13 16:20:31 -08:00
Harshit Jain
f33a7ca7d6
fix: as per request changes 2026-02-14 04:34:58 +05:30
Harshit Jain
37fc4a35ff
Update tests/test_litellm/proxy/management_endpoints/test_ui_sso.py
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
2026-02-14 04:27:13 +05:30
Harshit Jain
f87b12a2f7
fix grace period with better error handling on frontend and as per best practices 2026-02-14 04:18:08 +05:30
yuneng-jiang
b65cb646aa addressing comments 2026-02-13 13:59:49 -08:00
Ishaan Jaff
a06113ec82
feat: MCP OAuth2 client-side debug headers (#21151)
* fix: SCOPES on Atlassian issue

* feat: add MCPDebug class for client-side MCP OAuth2 debugging

* feat: inject MCP debug headers into streamable HTTP response path

* test: add unit tests for MCPDebug class

* fix: refactor MCPDebug - move all logic into class static methods

* fix: collapse server.py debug code to two-liner using MCPDebug methods

* test: add tests for resolve_auth_resolution and wrap_send_with_debug_headers

* docs: add MCP debug headers section to troubleshooting guide

* docs: add Debugging OAuth section to mcp_oauth.md

* docs: replace inline debug section with cross-link to mcp_oauth

* docs: extract UI troubleshooting into its own page

* docs: simplify troubleshoot.md to issue reporting only

* docs: add quick-start debug command to MCP troubleshoot page

* docs: restructure sidebar - UI, MCP, Performance, Issue Reporting
2026-02-13 12:55:47 -08:00
yuneng-jiang
938a920372 cost breakdown fix 2026-02-13 12:51:55 -08:00
Ishaan Jaffer
817de99177 UI build 2026-02-13 12:51:17 -08:00
Ishaan Jaff
40b290aa84
fix: MCP Gateway SCOPES on Atlassian issue (#21150)
* fix: SCOPES on Atlassian issue

* test: add regression tests for scopes=None in OAuth discovery endpoints
2026-02-13 12:43:56 -08:00
yuneng-jiang
71107b3fd0
Merge pull request #21149 from BerriAI/litellm_spendlogs_reset_custom_time
[Fix] UI - Spend Logs: Reset Filters Resets Custom Date Range
2026-02-13 12:30:45 -08:00
yuneng-jiang
eb144a44d0
Merge pull request #21144 from BerriAI/litellm_ui_fallbacks_10
[Refactor] UI - Fallbacks: Default Configurable to 10 Models
2026-02-13 12:30:35 -08:00
yuneng-jiang
82fcc39e6b reset custom time range when reset filters is clicked 2026-02-13 12:12:50 -08:00
Ishaan Jaff
568edde87a
docs: fix Claude Code MCP tutorial with correct config and URL patterns (#21145)
- Fix Atlassian config: use http transport and correct URL (/v1/mcp not /v1/sse)
- Fix URL pattern: use standard /mcp/<server_name> not legacy /<server_name>/mcp
- Add parameter breakdown table explaining each claude mcp add argument
- Add warning that server name in proxy config must match URL path
- Add ngrok step for OAuth callback accessibility
- Add ~/.claude.json config option alongside claude mcp add
- Fix auth header guidance: use x-litellm-api-key for OAuth servers
2026-02-13 12:12:38 -08:00
yuneng-jiang
6eeb9b109b
Merge pull request #21143 from BerriAI/litellm_model_id_search
[Feature] UI - Spend Logs: Sorting Columns
2026-02-13 11:53:16 -08:00
yuneng-jiang
724ab1c154
Update ui/litellm-dashboard/src/components/networking.tsx
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
2026-02-13 11:53:04 -08:00
Krish Dholakia
24b56a14eb
Guardrails - new Policy Templates (pre-configured guardrail combinations for specific use-cases) (#21025)
* feat(patterns.json): add australia specific pii patterns - tax file number, abn, medicare number

Improve PII detection for australian contexts

* feat(patterns.json): add iban + street address pattern detection

* feat: support policy templates on ui

allows admin to enable pre-configured guardrails

helps cover specific use-cases well

* feat: create missing guardrails, working policy templates

* feat: policy_templates.json

support hosted policy templates

allows others to contribute to the policy templates

* docs: document new policy templates

* fix: address greptile feedback

* fix: fix linting error
2026-02-13 11:53:02 -08:00
yuneng-jiang
25fccad127 Change default fallbacks to 10 models 2026-02-13 11:50:13 -08:00
yuneng-jiang
adc2859f0a sorting spend logs in ui 2026-02-13 11:33:41 -08:00
datzscaler
aab8edde67
fix(guardrails): Zscaler AI Guard bug fixes and support during post-call (#20801)
* fix(guardrails): fixed post-call issue with Zscaler guardrail and invalid headers. Added unittests

* fix(guardrails): Addressed greptil comments. Make policyid hanlding more clear

* fix(guardrails): Address greptile comment

* Update litellm/proxy/guardrails/guardrail_hooks/zscaler_ai_guard/zscaler_ai_guard.py

Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>

---------

Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
2026-02-13 11:28:14 -08:00
Sameer Kankute
066e694f5e
Merge pull request #21110 from BerriAI/litellm_litellm_anthropic_remote_url3
Add support for remote URL fetching for anthropic beta header mapping
2026-02-14 00:30:51 +05:30
Ryan Crabbe
9c87484b07 fix: preserve auto-generated id when provider returns falsy id
Address review comment: fall back to model_response_object.id instead
of overwriting with None/empty string. Add parametrized test for the
edge case.
2026-02-13 10:13:45 -08:00
Ryan Crabbe
ecb04187b5 perf: optimize user_api_key_auth hot path (Rounds 1-3, 5-6)
- Fast-path early return in check_api_key_for_custom_headers_or_pass_through_endpoints
- startswith(tuple) replaces for-loop substring matching (also more correct)
- Pre-compute _PUBLIC_ROUTES and MAPPED_PASS_THROUGH_PREFIXES at module level
- Deduplicate get_request_route call by passing route as parameter
- Cache dict(request.headers) on request.state per-request
- Remove redundant elif branch from get_api_key (handled by later function)
- Fix: isinstance(request.headers, dict) was always False for Starlette Headers,
  silently breaking custom header extraction for pass-through endpoints
2026-02-13 10:00:38 -08:00
Ishaan Jaff
a1b7261502
Remove exposed API key from example configuration in mcp_server_manager.py (#21129)
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Ishaan Jaff <ishaan-jaff@users.noreply.github.com>
2026-02-13 09:52:38 -08:00
Julio Quinteros Pro
5bcd3b53c9 fix: import LiteLLM_ObjectPermissionTable from _types instead of proxy_server
Per maintainer feedback, FastAPI should always be available in proxy code.
The issue was that MCP tests were importing from proxy_server unnecessarily,
pulling in all proxy dependencies including policy_resolve_endpoints.

Fix:
- Revert policy_resolve_endpoints.py to use direct FastAPI imports
- Update MCP tests to import LiteLLM_ObjectPermissionTable from litellm.proxy._types
  instead of litellm.proxy.proxy_server

This avoids importing the entire proxy_server module with all its dependencies
when tests only need specific types.

Addresses: https://github.com/BerriAI/litellm/pull/21075/changes#r2802201174
2026-02-13 14:46:03 -03:00
Julio Quinteros Pro
1fffeb953c fix: restore if TYPE_CHECKING block for proper FastAPI import handling
Greptile's previous suggestion accidentally removed the if TYPE_CHECKING block,
leaving an orphan else statement that caused a syntax error.
This commit restores the proper structure.
2026-02-13 14:46:03 -03:00
jquinter
ca9cdea8a7 Update litellm/proxy/policy_engine/policy_resolve_endpoints.py
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
2026-02-13 14:46:02 -03:00
Julio Quinteros Pro
e6385a6c83 fix: make policy_resolve_endpoints importable without FastAPI
Adds try/except around FastAPI imports with fallback mock classes.
This allows the module to be imported in test environments where
proxy dependencies (FastAPI) may not be installed.

Fixes NameError when MCP tests try to import from proxy_server which
imports from this module:
- NameError: name 'APIRouter' is not defined
- NameError: name 'Depends' is not defined
- NameError: name 'HTTPException' is not defined
- NameError: name 'Query' is not defined
2026-02-13 14:46:02 -03:00
Julio Quinteros Pro
64424d00c1 fix: use None instead of Reasoning() for reasoning parameter
The ResponsesAPIResponse.reasoning field expects Optional[Dict[str, Any]],
not a Reasoning object. Passing an empty Reasoning() object causes a mypy type error.

Fix: Pass None instead of Reasoning() since the field is optional and we're
not providing any reasoning data.

Fixes mypy error:
Argument "reasoning" to "ResponsesAPIResponse" has incompatible type
"Reasoning"; expected "dict[str, Any] | None"
2026-02-13 14:30:05 -03:00
Aidan Kovacic
7a164ba2cc
fix: shared health check serialization (#21119)
* Add BaseModel serialization to safe_json_dumps

* Use safe_dumps in shared health check caching

* remove redundant seen.add
2026-02-13 09:15:43 -08:00
yuneng-jiang
5ddba48df9
Merge pull request #21088 from BerriAI/litellm_spend_rebase
[Fix] Spend Management Tests
2026-02-13 09:11:42 -08:00
Sameer Kankute
d4dbb460cc Make tests run with local beta header mapping json 2026-02-13 22:31:42 +05:30
Sameer Kankute
aaf5938864
Merge pull request #21089 from BerriAI/litellm_vector_store_endpoints
Add target_model_names for vector store endpoints
2026-02-13 22:18:29 +05:30
jquinter
93e8127562
chore: remove redundant test-complete job from test workflow (#21106)
The test-complete aggregate job adds no value as GitHub Actions
already provides visibility into matrix job results.

Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>
2026-02-13 08:39:00 -08:00
jquinter
1e0b1c8334
fix: remove duplicate PerplexityResponsesConfig from LLM_CONFIG_NAMES (#21105)
Removes duplicate entry of PerplexityResponsesConfig that was causing
Ruff linting error F601 (dictionary key literal repeated).

Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>
2026-02-13 08:38:50 -08:00
jquinter
d32b654d6b
fix: regenerate poetry.lock to sync with pyproject.toml (#21108)
poetry.lock was out of sync with pyproject.toml, causing CI failures
across all PRs with "Run `poetry lock` to fix the lock file" error.

Regenerated the lock file to resolve the sync issue.

Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>
2026-02-13 08:38:29 -08:00