* fix: remove leading space from license public_key.pem
PEM must begin with -----BEGIN; a leading ASCII space breaks
cryptography.load_pem_public_key on older cryptography (e.g. 41.x),
causing OpenSSL no start line / deserialize errors.
Made-with: Cursor
* test: assert license public_key.pem loads as valid PEM
Regression guard for leading whitespace before -----BEGIN, which breaks
load_pem_public_key on older cryptography (e.g. 41.x).
Made-with: Cursor
Fixes 'LLM Provider NOT provided' errors when models are configured with
custom_llm_provider but model names lack provider prefix (e.g., 'gpt-4.1-mini'
instead of 'azure/gpt-4.1-mini').
Changes:
- Router now passes deployment's custom_llm_provider to get_llm_provider()
- Fixes 6 code paths: file creation, file content, batch operations, vector store
- Adds regression tests for file creation and file content operations
Made-with: Cursor
* fix(vertex_ai): normalize Gemini finish_reason enum through map_finish_reason in streaming handler
In the legacy vertex_ai SDK streaming path, the raw Gemini finish_reason enum name (e.g. "STOP", "MAX_TOKENS") was stored directly into self.received_finish_reason without being mapped to OpenAI-compatible values. The finish_reason_handler then compared against lowercase "stop", causing the case mismatch to prevent the tool_call override from ever firing. This fix applies map_finish_reason() so all Gemini enum names are normalized before storage.Refactor finish reason handling to use map_finish_reason function.
* refactor: use module-level map_finish_reason import; drop redundant inline import
map_finish_reason is already imported at module scope (line 49) via `from .core_helpers import map_finish_reason, process_response_headers`. The inline import added in the previous commit was redundant. Addressed Greptile review feedback.Removed unnecessary import of map_finish_reason from core_helpers.
* test: add unit tests for Gemini legacy vertex finish_reason normalisation
Added tests to ensure finish_reason normalization for Gemini legacy vertex tool calls and stop reasons.
Anthropic/Claude Code use input_schema.type "custom"; Bedrock rejects it.
- Add normalize_json_schema_custom_types_to_object and use it for Invoke,
chat invoke, and _bedrock_tools_pt (Anthropic input_schema + OpenAI params).
- Coerce invalid root types to object for Converse toolSpec.
- Tests for invoke transform, converse _bedrock_tools_pt, and unit helper.
Made-with: Cursor
* feat(triton): add embedding usage estimation for self-hosted responses
Populate Triton embedding usage from request input using token counting with a safe fallback so cost/observability flows work even when provider usage is missing.
Made-with: Cursor
* fix(triton): sum per-input embedding token counts for batches
Joining batch strings with newlines before token_counter added spurious
tokens. Count each input separately and sum, matching OpenAI-style usage.
Made-with: Cursor
Using setdefault('litellm_metadata', {}) unconditionally created an empty
litellm_metadata key for chat completions and embeddings. This caused
_get_metadata_variable_name_from_kwargs to return 'litellm_metadata' instead
of 'metadata', so tag-based routing looked for tags in the wrong dict and
ignored all tag filters.
Fix: only set the encrypted_content_affinity_enabled flag when litellm_metadata
already exists (Responses API path). Chat completions and embeddings never have
this key, so nothing is created and tag routing works correctly.
Adds `litellm.route_all_chat_openai_to_responses` (env: `LITELLM_ROUTE_ALL_CHAT_OPENAI_TO_RESPONSES`) to route all OpenAI /chat/completions requests through the Responses API bridge. Also fixes reasoning param dict passthrough in completion transformation.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Allows openai/-prefixed models with a custom api_base pointing to a
third-party OpenAI-compatible provider to opt-in to the
/responses → /chat/completions bridge, rather than forwarding requests
natively to /v1/responses (which may not be supported by the provider).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Defense-in-depth: warn instead of hard-fail for legacy servers
- Move os import to module level in _types.py
- Document args residual risk in allowlist comment
- Add UpdateMCPServerRequest allowlist test
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add command allowlist for MCP stdio transport to prevent RCE via
/mcp-rest/test/* endpoints. Restrict test endpoints to PROXY_ADMIN
role. Fix docker/README.md MASTER_KEY -> LITELLM_MASTER_KEY.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Allow JWT tokens matching routing_overrides to use OAuth2 introspection without enabling global OAuth2 while keeping OAuth2 routing limited to LLM/info routes. Add regression coverage for management-route boundary and tighten opaque-token assertions; update docs to reflect selective-mode route scope.
Made-with: Cursor
PR #25258 changed _cleanup_stale_managed_objects from update_many to
execute_raw via _expire_stale_rows, but the tests were not updated.
The tests now mock _expire_stale_rows on the instance and assert
update_many calls only for job completion, not stale cleanup.
- team-admin: assert Admin Settings is not visible (role-specific check)
- proxy-admin: use users[Role.ProxyAdmin].password from constants instead of duplicating the env var fallback inline
* fix: batch-limit stale managed object cleanup to prevent 300K row UPDATE (#25257)
* Add STALE_OBJECT_CLEANUP_BATCH_SIZE constant
Configurable batch limit (default 1000) for stale managed object cleanup,
preventing unbounded UPDATE queries from hitting 300K+ rows at once.
* Batch-limit stale managed object cleanup with single bounded SQL query
Two fixes to _cleanup_stale_managed_objects:
1. Replace unbounded update_many with a single execute_raw using a
subquery LIMIT, capping each poll cycle to STALE_OBJECT_CLEANUP_BATCH_SIZE
rows. Zero rows loaded into Python memory — everything stays in Postgres.
Uses the same PostgreSQL raw-SQL pattern as spend_log_cleanup.py
(the proxy requires PostgreSQL per schema.prisma).
2. Extract _expire_stale_rows as a separate method for testability.
Keeps the file_purpose='response' filter to avoid incorrectly expiring
long-running batch or fine-tune jobs that legitimately exceed the
staleness cutoff.
* docs: add STALE_OBJECT_CLEANUP_BATCH_SIZE to env vars reference
* test: remove deprecated embed-english-v2.0 cohere embedding tests
Adds a new endpoint to bulk-update team_member_permissions across
teams. Supports apply_to_all_teams (with cursor-based pagination)
or a specific list of team_ids. Merges new permissions into each
team's existing set rather than overwriting.
Also fixes test isolation bug in test_get_prompt_info_by_base_id
where leaked prisma_client state from other tests caused a
TypeError on await.
The proxy_e2e_azure_batches_tests workflow is consistently flaky and
does not provide reliable signal on whether changes break anything.
Remove the workflow from both CircleCI and GitHub Actions, along with
the test directory it exclusively used.
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat: multiple concurrent budget windows per API key and team (#24883)
* feat(proxy): add BudgetLimitEntry type and wire budget_limits into key/team models
* feat(schema): add budget_limits Json column to VerificationToken and TeamTable
* feat(migrations): add migration for budget_limits column on keys and teams
* feat(keys): initialize budget_limits windows with reset_at on key create/update
* feat(teams): initialize budget_limits windows with reset_at on team create/update
* feat(auth): add _virtual_key_multi_budget_check and _team_multi_budget_check
* feat(auth): call multi-budget checks from common_checks for keys and teams
* feat(proxy): increment per-window Redis spend counters after each request
* feat(budget): reset individual budget windows on schedule via reset_budget_job
* feat(ui): add hourly option to BudgetDurationDropdown
* feat(ui): add budget_limits field to KeyResponse type
* feat(ui): add Budget Windows editor to key edit view
* feat(ui): add Budget Windows editor to create key form
* fix(proxy): strip budget_limits=None before Prisma upsert to fix login 500
Prisma rejects nullable JSON fields (Json? without @default) when passed as
Python None — it needs the field omitted entirely so the DB stores NULL via
the column's nullable constraint. This was breaking /v2/login because the UI
session key creation path hit the upsert with budget_limits=None.
* ui(key-edit): use antd InputNumber+Button for budget windows, add reset hints
* ui(create-key): use antd InputNumber+Button for budget windows, add reset hints
* docs(users): add multiple budget windows section with API + dashboard walkthrough
* fix: BudgetExceededError returns HTTP 429 instead of 400
- Add status_code=429 to BudgetExceededError class
- auth_exception_handler hardcoded code=400 → code=429
* fix: no-op else branch in multi-budget auth checks causes KeyError
- BudgetLimitEntry objects must be coerced via model_dump() not left as-is
- Move _virtual_key_multi_budget_check into common_checks (was asymmetric
with _team_multi_budget_check which already lived there)
* fix: len() on JSON string returns char count not window count
Guard with isinstance check + json.loads() before iterating per-window
Redis counters in increment_spend_counters
* fix: silent except:pass hides Redis reset failures in reset_budget_windows
Log Redis counter reset failures as warnings so they are observable
* test: add unit tests for multi-budget window enforcement
5 tests covering: no budget_limits passes, under budget passes,
over hourly window raises 429, over monthly window raises 429,
BudgetLimitEntry objects coerced without KeyError
* fix: key per-window counters stable across reorders (duration key, not index)
* fix: team+key per-window spend increments use duration key, not index
* fix: budget window reset uses duration key; log failures instead of swallowing
* refactor: extract BudgetWindowsEditor to shared component
* refactor: key_edit_view imports BudgetWindowsEditor from shared component
* refactor: create_key_button imports BudgetWindowsEditor from shared component
---------
Co-authored-by: Ishaan Jaffer <ishaanjaffer0324@gmail.com>
* fix(reset_budget_job): extract _reset_expired_window helper to fix PLR0915 too many statements
* feat(skills): Skills Registry & Hub — register skills, browse in AI Hub, public skill hub (#25118)
* feat(skills): add domain and namespace fields to plugin types
* feat(skills): store and return domain/namespace inside manifest_json
* feat(skills): add /public/skill_hub endpoint for unauthenticated access
* feat(skills): whitelist /public/skill_hub from auth requirements
* feat(skills): add domain, namespace to Plugin and RegisterPluginRequest types
* feat(skills): smart URL parser — paste github URL, auto-detect source type and name
* feat(skills): replace enable toggle with Public badge, make rows clickable
* feat(skills): add skill detail view with Overview and How to Use tabs
* feat(skills): add MakeSkillPublicForm modal for publishing skills to the hub
* feat(skills): rename panel to Skills, wire in skill detail view on row click
* feat(skills): add skill hub table columns — name, description, domain, source, status
* feat(skills): add SkillHubDashboard with stats row, domain dropdown filter, and table
* feat(skills): add Skill Hub tab to AI Hub with Select Skills to Make Public button
* feat(skills): move Skills to top-level nav item directly under MCP Servers
* feat(skills): add skillHubPublicCall and NEXT_PUBLIC_BASE_URL support
* feat(skills): add Skill Hub tab to public AI Hub page
* feat(skills): add skills page routing in main app router
* feat(skills): add /skills page route
* chore: update package-lock after npm install
* docs(skills): add Skills Gateway doc page with mermaid architecture diagram
* docs(skills): add Skills Gateway to sidebar under Agent & MCP Gateway
* docs(skills): add loom walkthrough video to Skills Gateway doc
* chore: fixes
---------
Co-authored-by: Ishaan Jaffer <ishaanjaffer0324@gmail.com>
Co-authored-by: Yuneng Jiang <yuneng@berri.ai>
* fix(bedrock): strip [1m]/[200k] context window suffixes before cost lookup
* test(bedrock): add test for [1m] context window suffix stripping in cost lookup
* schema: add allowed_models to BudgetTable, default_team_member_models to TeamTable
* migration: add allowed_models and default_team_member_models columns
* types: add allowed_models to TeamMemberAddRequest, TeamMemberUpdateRequest, UpdateTeamRequest
* utils: add allowed_models param to add_new_member, persist to budget table
* common_utils: add allowed_models to _upsert_budget_and_membership
* team endpoints: seed allowed_models on member_add, persist on member_update and team/update
* auth: enforce per-member allowed_models at request time
* networking: add allowed_models to Member type and teamMemberUpdateCall
* TeamMemberTab: add Model Scope column showing per-member allowed_models
* EditMembership: add Allowed Models multi-select field
* TeamInfo: add default_team_member_models field in Settings tab
* chore: sync schema.prisma copies from root
* fix(team_member_update): update existing budget in-place instead of creating new one
When a member already has a budget_id, patch only the fields the caller
provided rather than always creating a fresh budget record. The old
code ignored existing_budget_id entirely, so updating only allowed_models
silently dropped the stored max_budget / tpm_limit / rpm_limit values.
* fix(auth): pass llm_router to _check_team_member_model_access
Without the router, _can_object_call_model cannot resolve wildcard model
names (e.g. openai/*) or access-group names in allowed_models, causing
legitimate requests to be denied. Thread the existing llm_router from
_run_common_checks through to the new member-scope check.
* feat(ui): add Team Member Settings accordion to Create Team modal
Groups default_team_member_models, member budget/key duration, and
tpm/rpm defaults into a single collapsible section. The model picker
is filtered to only show the models selected for the team, and the
copy distinguishes it from the team-level Models field.
* feat(ui): consolidate Team Member Settings into accordion in edit team form
Moves default_team_member_models + per-member budget/key/tpm/rpm fields
into a collapsible "Team Member Settings" panel. Keeps the top-level
form focused on team-wide settings (team models, team budget, tpm/rpm).
* fix(ui): use tremor Accordion for Team Member Settings in edit team form
* fix(ui): move Team Member Settings accordion above budget fields in Create Team
* chore: fixes
---------
Co-authored-by: Ishaan Jaffer <ishaanjaffer0324@gmail.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Yuneng Jiang <yuneng@berri.ai>
Extend LATENCY_BUCKETS beyond 5 minutes so request/LLM latency metrics
can distinguish long runs up to the typical default LLM request timeout.
Made-with: Cursor
* fix(presidio): use correct text positions in anonymize_text (#24160)
The Presidio anonymizer endpoint returns items with start/end positions
that reference the *anonymized output* text, not the original input.
anonymize_text() was applying these positions to the original text,
causing garbled output with remnants of un-masked PII data.
When output_parse_pii is False, return redacted_text["text"] directly
from the anonymizer response instead of manually splicing.
When output_parse_pii is True, use analyze_results positions (which
correctly reference the original text) to build numbered replacement
tokens and the pii_tokens mapping.
* address review: remove dead code, fix token numbering order
- Remove unused `anon_item_by_entity` dict (Greptile P2)
- Number tokens left-to-right (<PERSON_1> first in text, not last)
- Add assertion for token numbering order in test
## Problem
When `get_cache_key(**kwargs)` is called with kwargs that already
contains `preset_cache_key` (which can happen when cache key is
recomputed in certain code paths), the call to
`_set_preset_cache_key_in_kwargs()` fails with:
```
TypeError: _set_preset_cache_key_in_kwargs() got multiple values
for keyword argument 'preset_cache_key'
```
This is because `preset_cache_key` is passed both explicitly:
```python
self._set_preset_cache_key_in_kwargs(
preset_cache_key=hashed_cache_key, **kwargs
)
```
And implicitly via `**kwargs` unpacking when `kwargs["preset_cache_key"]`
exists.
## Solution
Filter out `preset_cache_key` from kwargs before passing to
`_set_preset_cache_key_in_kwargs()`:
```python
kwargs_for_preset = {k: v for k, v in kwargs.items() if k != "preset_cache_key"}
self._set_preset_cache_key_in_kwargs(
preset_cache_key=hashed_cache_key, **kwargs_for_preset
)
```
## Testing
Added unit tests covering:
- kwargs with existing preset_cache_key (the bug case)
- kwargs without preset_cache_key (regression test)
- Verification that preset_cache_key is correctly set in litellm_params
Fixes#25081.
is_tool_name_prefixed() checked for the presence of MCP_TOOL_PREFIX_SEPARATOR
(default '-') anywhere in the tool name. Any non-MCP tool whose name
contains a hyphen (e.g. 'text-to-speech', 'code-review') was silently
misclassified as an MCP-prefixed tool. When the semantic tool filter is
enabled, these tools would be routed through semantic matching and
potentially dropped.
Fix: accept an optional known_server_prefixes set. When supplied, the
function extracts the candidate prefix (text before the first separator)
and checks it against the normalised set of registered server prefixes.
Only a genuine match returns True. Without the set, legacy behaviour is
preserved for backward compatibility.
Updated _get_mcp_server_from_tool_name() to build the prefix set from
the live registry and pass it through.
9 new tests.
Co-authored-by: d 🔹 <258577966+voidborne-d@users.noreply.github.com>
* added support for metadata (#24261)
* added support for metadata
* fix: PR review - meta truthiness, BlobResourceContents mimeType, add Blob+empty meta tests
Made-with: Cursor
* pyproject to .25
* feat(teams): resolve access group models/MCPs/agents in team endpoints
Add access_group_models, access_group_mcp_server_ids, and
access_group_agent_ids to /team/info and /v2/team/list responses.
These fields contain resources inherited from access groups, kept
separate from direct assignments so the UI can distinguish the source.
Backend: _resolve_access_group_resources() helper resolves access
group resources via existing _get_*_from_access_groups() functions.
UI: Teams table and detail view show direct models as blue badges
and access-group-sourced models as green badges.
* perf(teams): single-pass access group resolution + asyncio.gather in list endpoint
- Fetch each access group object once and extract all 3 resource fields
in a single pass instead of 3 separate calls (3N → N lookups)
- Use asyncio.gather to resolve access groups across teams concurrently
in list_team_v2 instead of sequential awaits
- Add 5 unit tests for _resolve_access_group_resources
* docs: add default_team_params to config reference and update examples
- Add default_team_params to litellm_settings reference table in
config_settings.md with all sub-fields documented
- Update self_serve.md and msft_sso.md examples to include
team_member_permissions, tpm_limit, and rpm_limit
- Fix misleading comment that implied default_team_params only applies
to SSO auto-created teams — it applies to all /team/new calls
* docs: clarify that models sub-field only applies to SSO auto-created teams
* fix: lazy import get_access_object to break cyclic import + short-circuit all-proxy-models display
- Remove get_access_object from module-level import in team_endpoints.py
and use a lazy _get_access_object wrapper to avoid cyclic dependency
- Add _prisma_client is None early-exit guard in _resolve_access_group_resources
- Short-circuit UI to show "All Proxy Models" when team.models is empty
or contains "all-proxy-models", skipping access group model resolution
* add: making organizations a select instead of read only badges
* fix(ui): only send organization_id when changed and use raw initial value
* fix(ui): add paginated team search to usage page filter
Replace the static team dropdown on the usage page with a new
TeamMultiSelect component that uses the paginated v2/team/list
endpoint with debounced server-side search and infinite scroll.
* fix(ui): fix imports and update placeholder for team multi select
* fix(ui): wire team_id filter to key alias dropdown on Virtual Keys tab
The Key Alias dropdown on the Virtual Keys page was showing aliases from
all teams regardless of which team was selected. The team_id was never
passed through the frontend chain to the backend /key/aliases endpoint.
- Backend: add optional team_id query param to /key/aliases endpoint
- networking.tsx: add team_id param to keyAliasesCall
- useKeyAliases: accept and forward team_id to API call and query key
- filter.tsx: pass allFilters context to custom filter components
- PaginatedKeyAliasSelect: read Team ID from allFilters and pass to hook
* fix(tests): correct mock targets in TestResolveAccessGroupResources
Three tests were patching the non-existent `get_access_object` instead
of `_get_access_object` (the lazy-import wrapper), causing AttributeError.
Also added missing `prisma_client` mock so tests get past the early-exit
guard and actually exercise the resolution logic.
* fix: use direct attribute access with or [] fallback in _resolve_access_group_resources
Replace getattr(ag, "field", []) with ag.field or [] for cleaner
access and safe handling if a field is None.
* fix(ui): remove model source legend from team detail view
The blue/green color distinction is self-explanatory; the legend added
visual clutter without providing enough value.
* fix(ui): add missing access_group fields to TeamData.team_info type
The TeamData interface was missing access_group_models,
access_group_mcp_server_ids, and access_group_agent_ids fields,
causing a TypeScript build failure.
* perf(teams): batch-fetch access groups in single DB query
Replace per-ID _resolve_access_group_resources loop with a single
find_many call that deduplicates IDs across all teams. Removes the
N+1 query pattern on cold cache for the team list endpoint.
* refactor(proxy): extract helpers to fix PLR0915 violations
Extract `_apply_non_admin_alias_scope` from `key_aliases`,
`_resolve_team_access_group_resources` from `team_info`, and
`_enforce_list_team_v2_access` from `list_team_v2` to bring each
function under ruff's 50-statement limit. No behavior changes.
* test(ui): update tests to match new team_id / access-group signatures
- useKeyAliases, PaginatedKeyAliasSelect: add trailing `undefined` to
spy matchers for the new `team_id` param on `useInfiniteKeyAliases`
and `keyAliasesCall`.
- EntityUsage: mock new `TeamMultiSelect` child so QueryClientProvider
is not required for team-entity tests.
- ModelsCell: replace the overflow-accordion test with one that
verifies the new collapse-on-`all-proxy-models` behavior (no
accordion, single badge).
* fix(ui): send null (not '') for cleared organization_id on team update
AntD <Select allowClear> returns undefined when the user clears the
selection. Coalescing to "" caused the team-update payload to carry
organization_id: "" instead of null, relying on the backend to coerce
it. Send null directly so the intent is explicit at the source.
* poetry
* chore: regen poetry.lock for litellm-proxy-extras 0.4.64 bump
* chore: update Next.js build artifacts (2026-04-04 17:55 UTC, node v22.16.0)
---------
Co-authored-by: shivam <shivam@uni.minerva.edu>
Co-authored-by: Ryan Crabbe <ryan@berri.ai>
Co-authored-by: yuneng-jiang <yuneng@berri.ai>
* Tag query fix (#25094)
* feat(tag-spend): implement separate scheduler job for daily tag spend updates
* fix(docker): add g++ to build dependencies in Dockerfile
* initial test cases. TODO: check scheduler init and test cases in proxy_server related to it
* resolved QPS issue when redis transaction buffer is enabled
* resolving circular import error flagged by greptile
* fix(mypy): use Optional[str] for api_base in PydanticAI provider to match superclass signature
---------
Co-authored-by: Shivam Rawat <shivam@berri.ai>
Co-authored-by: shivam <shivam@uni.minerva.edu>
Co-authored-by: Ryan Crabbe <ryan@berri.ai>
Co-authored-by: yuneng-jiang <yuneng@berri.ai>
Co-authored-by: Harish <harishgokul01@gmail.com>
Co-authored-by: Ishaan Jaffer <ishaan@berri.ai>
* Litellm ishaan march23 - MCP Toolsets + GCP Caching fix (#25146)
* feat(mcp): MCP Toolsets — curated tool subsets from one or more MCP servers (#24335)
* feat(mcp): add LiteLLM_MCPToolsetTable and mcp_toolsets to ObjectPermissionTable
* feat(mcp): add prisma migration for MCPToolset table
* feat(mcp): add MCPToolset Python types
* feat(mcp): add toolset_db.py with CRUD helpers for MCPToolset
* feat(mcp): add toolset CRUD endpoints to mcp_management_endpoints
* fix(mcp): skip allow_all_keys servers when explicit mcp_servers permission is set (toolset scope fix)
* feat(mcp): add _apply_toolset_scope and toolset route handling in server.py
* fix(mcp): resolve toolset names in responses API before fetching tools
* feat(mcp): add mcp_toolsets field to LiteLLM_ObjectPermissionTable type
* feat(mcp): register LiteLLM_MCPToolsetTable in prisma client initialization
* feat(mcp): validate mcp_toolsets in key-vs-team permission check
* feat(mcp): register toolset routes in proxy_server.py
* feat(mcp): add MCPToolset and MCPToolsetTool TypeScript types
* feat(mcp): add fetchMCPToolsets, createMCPToolset, updateMCPToolset, deleteMCPToolset API functions
* feat(mcp): add useMCPToolsets React Query hook
* feat(mcp): add toolsets (purple) as third option type in MCPServerSelector
* feat(mcp): extract toolsets from combined MCP field in key form
* feat(mcp): extract toolsets from combined MCP field in team form
* feat(mcp): show toolsets section in MCPServerPermissions read view
* feat(mcp): pass mcp_toolsets through object_permissions_view
* feat(mcp): add MCPToolsetsTab component for creating and managing toolsets
* feat(mcp): add Toolsets tab to mcp_servers.tsx
* feat(mcp): pass mcpToolsets to playground chat and responses API calls
* feat(mcp): generate correct server_url for toolsets in playground API calls
* docs(mcp): add MCP Toolsets documentation
* docs(mcp): add mcp_toolsets to sidebar
* fix(mcp): replace x-mcp-toolset-id header with ContextVar to prevent client forgery
* fix(mcp): use ContextVar + StreamingResponse for toolset MCP routes (fixes SSE streaming)
* fix(mcp): cache toolset permission lookups to avoid per-request DB calls
* test(mcp): add tests for toolset scope enforcement, ContextVar isolation, and access control
* fix(mcp): cache toolset name lookups in MCPServerManager to avoid per-request DB calls
* fix(mcp): prevent body_iter deadlock + use cached toolset lookup in responses API
- _stream_mcp_asgi_response: add done callback to handler_task that puts
the EOF sentinel on body_queue when the task exits, preventing body_iter
from hanging forever if the handler raises after headers are sent.
- litellm_proxy_mcp_handler: replace raw get_mcp_toolset_by_name() DB call
with global_mcp_server_manager.get_toolset_by_name_cached() so toolset
resolution uses the 60s TTL cache added for this purpose instead of
hitting the DB on every responses-API request.
* fix(mcp): toolset access control, asyncio fix, and real unit tests
- server.py: _apply_toolset_scope now enforces that non-admin keys must
have the requested toolset_id in their mcp_toolsets grant list;
admin keys always bypass the check.
- mcp_management_endpoints.py: three access-control fixes:
* fetch_mcp_toolsets: non-admin keys with mcp_toolsets=None now
return [] instead of all toolsets (only admins get 'all' when
the field is absent)
* fetch_mcp_toolset: non-admin keys that haven't been granted the
requested toolset_id now get 403 instead of the full result
* add_mcp_toolset: duplicate toolset_name now returns 409 Conflict
instead of an opaque 500
- proxy_server.py: use asyncio.get_running_loop() instead of
get_event_loop() inside an already-running coroutine (Python 3.10+).
- test_mcp_toolset_scope.py: replace four hollow tests that only
asserted local variable properties with real tests that call the
production fetch_mcp_toolsets() and handle_streamable_http_mcp()
functions with mocked dependencies.
* fix(mcp): add mcp_toolsets to ObjectPermissionBase, fix multi-toolset overwrite, fix delete 404, allow standalone key toolsets
* fix(mcp): add auth check on toolset resolution in responses API; union mcp_servers in _merge_toolset_permissions
* fix(mcp): handle RecordNotFoundError in update_mcp_toolset; union direct servers with toolset servers
* fix(mcp): use _user_has_admin_view; deny None mcp_toolsets for non-admin; use direct RecordNotFoundError import; fix docstring
* fix(mcp): add @default(now()) to MCPToolsetTable.updated_at; fix test for non-admin toolset access
* fix: use UniqueViolationError import; guard _ensure_eof for error/cancel only
* fix(mcp): preserve mcp_access_groups in toolset scope, use shared Redis cache for toolset perms
- Remove mcp_access_groups=[] from _apply_toolset_scope (server.py) and the
responses API toolset path (litellm_proxy_mcp_handler.py). A key's access-group
grants remain valid even when the request is scoped to a single toolset; clearing
them silently revoked legitimate entitlements.
- Switch resolve_toolset_tool_permissions and get_toolset_by_name_cached to use
user_api_key_cache (Redis-backed DualCache in production) instead of per-instance
in-memory dicts. Cache entries are now shared across workers, eliminating the
per-worker stale-toolset-permission window flagged as a P1 by Greptile.
- Use union merge (set union of tool names per server) when applying toolset
permissions in the responses API path so direct-server tool restrictions are not
overwritten by toolset permissions.
* fix(mcp): return 404 when edit_mcp_toolset target does not exist
* fix(mcp): align mcp_toolsets default to None in LiteLLM_ObjectPermissionTable
* fix(mcp): admin toolset visibility, in-place tool name mutation, test helper coercion
* fix(mcp): treat None/[] team mcp_toolsets as no restriction in key validation
* fix(mcp): allow_all_keys backward compat, blocked_tools API write-path, efficient startup query
* fix(mcp): use _mcp_active_toolset_id ContextVar to detect toolset scope, avoiding DB-default false-positive
* fix(mcp): remove dead toolset cache stubs, log invalidation failures, align schema updated_at defaults
* fix(mcp): deserialise MCPToolset from Redis cache hit, replace fastapi import in test
* fix(mcp): evict name-cache on toolset mutation, 409 on rename conflict, warning-level list errors
* fix(redis): regenerate GCP IAM token per connection for async cluster (#24426)
* fix(redis): regenerate GCP IAM token per connection for async cluster clients
Async RedisCluster was generating the IAM token once at startup and
storing it as a static password. After the 1-hour GCP token TTL, any
new connection (including to newly-discovered cluster nodes) would fail
to authenticate.
Fix: introduce GCPIAMCredentialProvider that implements redis-py's
CredentialProvider protocol. It calls _generate_gcp_iam_access_token()
on every new connection, matching what the sync redis_connect_func
already does. async_redis.RedisCluster accepts a credential_provider
kwarg which is invoked per-connection.
* refactor(redis): move GCPIAMCredentialProvider to its own file
Extract GCPIAMCredentialProvider and _generate_gcp_iam_access_token
into litellm/_redis_credential_provider.py. _redis.py imports them
from there, keeping the public API unchanged.
* fix: address Greptile review issues
- GCPIAMCredentialProvider now inherits from redis.credentials.CredentialProvider
so redis-py's async path calls get_credentials_async() properly
- move _redis_credential_provider import to top of _redis.py (PEP 8)
- remove dead else-branch that silently no-oped (gcp_service_account from
redis_kwargs.get() was always None since it's popped by _get_redis_client_logic)
- remove mid-function 'from litellm import get_secret_str' inline import
- remove unused 'call' import from test_redis.py
* chore: retrigger CI/review
* chore: sync schema.prisma copies from root
* chore: sync schema.prisma copies from root
* fix(proxy_server): use bounded asyncio.Queue with maxsize to prevent unbounded growth
* fix(a2a/pydantic_ai): make api_base Optional to match base class signature
* fix(a2a/pydantic_ai): make api_base Optional in handler and guard against None
* fix(mcp): remove unused get_all_mcp_servers import
* fix(mcp): remove unused MCPToolset import
* refactor(mcp): extract toolset permission logic to reduce statement count below PLR0915 limit
* fix(tests): update reload_servers_from_database tests to mock prisma directly
---------
Co-authored-by: Ishaan Jaff <ishaanjaffer0324@gmail.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* fix(toolset_db): lazy-import prisma to avoid ImportError when prisma not installed
* fix(tests): update UI tests for toolset tab and updated empty state text
* fix(tests): add get_mcp_server_by_name to fake_manager stub
---------
Co-authored-by: Ishaan Jaff <ishaanjaffer0324@gmail.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Bug 1: internal users hit route-level 403 on /guardrails/submissions.
The route wasn't in self_managed_routes, so the route allowlist rejected
non-admin callers before our endpoint's team-scoping ran. Added
/guardrails/submissions and /guardrails/submissions/{guardrail_id} to
self_managed_routes.
Bug 2: register_guardrail 403'd non-admins registering for teams in
their user.teams list. It used get_team_membership() which reads the
litellm_teammembership join table, but that row is only created when
the team has a budget (management_helpers/utils.py:225). Switched to
the _get_user_team_ids helper (reads user_obj.teams), making it
consistent with list_guardrail_submissions.
UI: moved the Test Playground tab inside the isAdmin conditional in
guardrails.tsx. Internal users now see only the Submitted Guardrails
tab; admins still see all four.
Tests: added coverage for non-admin register paths (cross-team allowed
and cross-team forbidden).
* feat(router): integrate allowed_fails_policy into health check failures (#24988)
* feat(router): integrate allowed_fails_policy into health check failures
Health check failures now increment the same per-deployment failure
counters used by allowed_fails_policy, so users can control how many
health check failures of each error type are required before a
deployment enters cooldown.
- ahealth_check() preserves the original exception in its return dict
- run_with_timeout() returns a litellm.Timeout on health check timeout
- _perform_health_check() propagates exceptions to unhealthy endpoints
- _write_health_state_to_router_cache() calls _set_cooldown_deployments
for each unhealthy endpoint that has an exception
- When allowed_fails_policy is set, the binary health check filter is
bypassed so cooldown is the sole routing exclusion mechanism
- Safety net: if all deployments are in cooldown with
enable_health_check_routing=True, the cooldown filter is bypassed
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat(router): add health_check_ignore_transient_errors flag
When enabled, health check failures with 429 (rate limit) or 408 (timeout)
status codes are skipped from the cooldown pipeline. These are transient
load issues, not broken deployments. Auth errors (401), 404, and 5xx errors
still increment counters and trigger cooldown as before.
Config (general_settings):
health_check_ignore_transient_errors: true
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(router): also exclude 429/408 from health state cache when ignore_transient_errors set
The previous fix only skipped cooldown counter increments. The health state
cache was still marking 429/408 endpoints as is_healthy=False, causing the
binary health check filter to exclude them from routing.
Now, when health_check_ignore_transient_errors=True, 429/408 endpoints are
also excluded from the unhealthy list passed to build_deployment_health_states(),
so the binary filter treats them as unaffected (not unhealthy).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* docs(router): add health check driven routing guide
New standalone page covering the full health check routing feature:
allowed_fails_policy integration, health_check_ignore_transient_errors,
architecture SVG, step-by-step setup, and gotchas (TTL, AllowedFails semantics).
Replaces the inline section in health.md with a link to the new page.
Added to the Routing & Load Balancing sidebar.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(health-check-routing): fix three CI failures
- Add "exception" to ILLEGAL_DISPLAY_PARAMS in health_check.py so the
exception object is stripped before the health endpoint serializes
results to JSON (fixes TypeError: 'URL' object is not iterable)
- Add allowed_fails_policy = None to FakeRouter stubs in
test_router_health_check_routing.py (fixes AttributeError)
- Add health_check_ignore_transient_errors to config_settings.md router
settings reference table (fixes documentation test)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* Fix litellm/tests/proxy_unit_tests/test_proxy_server.py
* fix(router): address greptile review comments
- Narrow cooldown safety-net bypass: only fires when allowed_fails_policy
is set (cooldown is health-check driven). Without a policy, cooldowns
are from real request failures and must not be bypassed.
- Restore cooldown deployments DEBUG log that was accidentally removed.
- Fix test_health TypeError: move exception extraction to a separate
exceptions_by_model_id dict returned alongside endpoints, so exception
objects never appear in the endpoint dicts that get JSON-serialized
by the /health response.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(health-check-routing): properly isolate exceptions from health response
Return exceptions_by_model_id as a separate third value from
_perform_health_check / perform_health_check so exception objects
(which contain non-JSON-serializable httpx URL types) never appear
in the endpoint dicts that get serialized by the /health response.
Callers updated: _health_endpoints.py, shared_health_check_manager.py,
proxy_server.py background loop. All use the exceptions dict only for
cooldown integration, not for display.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(shared-health-check): fix remaining 2-value return sites and update type annotation
* fix(health-check-routing): fix P0 cooldown integration never firing
The cooldown loop was reading endpoint.get("exception") which is always
None because exceptions are now returned via exceptions_by_model_id, not
stored in endpoint dicts. Fixed to use _exceptions.get(model_id).
Also fixes the transient-error filter to use _exceptions instead of
endpoint.get("exception"), and fixes all remaining 2-value return sites
in shared_health_check_manager.py. Tests updated to pass exceptions via
exceptions_by_model_id parameter instead of endpoint dicts.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(health-check-routing): fix P1 transient-error filter broken on cache hits
When SharedHealthCheckManager returns cached results, exceptions_by_model_id
is always {} so the transient-error filter defaulted to status 500 for all
endpoints, incorrectly marking 429/408 endpoints as unhealthy.
Fix: store integer exception_status on each unhealthy endpoint dict in
_perform_health_check. _get_endpoint_exception_status() uses the live
exception object when available (direct path) and falls back to the stored
integer (cache-hit path). The integer is JSON-serializable and survives
the shared cache round-trip.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(health-check-routing): gate cooldown loop behind allowed_fails_policy
Without the policy, cooldown is not the routing exclusion mechanism.
Firing _set_cooldown_deployments for all enable_health_check_routing users
was a backwards-incompatible change — 401s would immediately cooldown
deployments that the binary filter would have recovered on the next cycle.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* revert: undo allowed_fails_policy gate on cooldown loop
Cooldown integration via health checks is intentional for all
enable_health_check_routing users, not just those with allowed_fails_policy.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(docs+tests): fix health_check_ignore_transient_errors doc section and test coverage
- Move health_check_ignore_transient_errors from router_settings to
general_settings in config_settings.md (code reads it from general_settings)
- Remove duplicate enable_health_check_routing / health_check_staleness_threshold
entries that were incorrectly listed under router_settings
- Replace TestHealthCheckEndpointExceptionPropagation tests with ones that
exercise the real _perform_health_check code path via mocked ahealth_check,
verifying exceptions appear in exceptions_by_model_id and NOT in endpoint dicts
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(tests+docs): fix tuple unpacking and docs test failures
- Update test mocks that return (healthy, unhealthy) to return
(healthy, unhealthy, {}) to match the new 3-value signature
- Update test unpackings of perform_shared_health_check to use
healthy, unhealthy, _ = ...
- Add health_check_ignore_transient_errors to router_settings section
in config_settings.md (it is a Router constructor param, so the doc
test requires it there; it also lives in general_settings for proxy use)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* Fix CodeQL errors
* fix(tests): fix 2-value unpackings of _perform_health_check in test_health_check.py
* fix(tests): fix mock _perform_health_check returning 2-tuple instead of 3
* fix team routing
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: add distributed lock for key rotation job (#23364)
* fix: add distributed lock for key rotation job
* fix: address Greptile review feedback on key rotation lock (#23834)
* fix: address Greptile review feedback on key rotation lock
* fix req changes greptile
* feat(proxy): Optional on_error for guardrail pipeline (API / technical failures) (#24831)
* guardrails fallback
* docs
* docs: add LITELLM_KEY_ROTATION_LOCK_TTL_SECONDS to environment variables reference
* fix(mypy): accept Union[Dict, Any] in _get_deployment_order and use typed list to fix min() type error
* fix(mypy): use Optional[str] for api_base in PydanticAI provider to match superclass signature
---------
Co-authored-by: Sameer Kankute <sameer@berri.ai>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: Harshit Jain <48647625+Harshit28j@users.noreply.github.com>
Co-authored-by: Shivam Rawat <shivam@berri.ai>
Co-authored-by: yuneng-jiang <yuneng@berri.ai>
Backend:
- list_guardrail_submissions no longer 403s non-admins; it returns only
submissions whose team_id matches one of the caller's teams (via
get_user_object.teams). Admins still see all.
- Filtering by a team the caller is not in returns 403.
- Users with no team memberships get an empty list (no DB query).
- get_guardrail_submission applies the same scoping to single-item GETs.
Frontend:
- Remove admin-only bail-out in TeamGuardrailsTab.fetchSubmissions so
internal users actually load their team's submissions.
- Finish antd migration in guardrails.tsx: drop the last Tremor Button.
- Remove guardrailsList.length === 0 gate on the Test Playground tab;
the playground already renders a "No guardrails available" inline
empty state, which is more discoverable than a disabled tab.
Tests:
- Cover non-admin scoped access, empty teams, cross-team filter 403,
and per-submission GET scoping.
* fix(pricing): add unversioned vertex_ai/claude-haiku-4-5 entry
Missing unversioned entry causes cost tracking to return $0.00 for
all requests using vertex_ai/claude-haiku-4-5. All other Vertex AI
Claude models have both versioned and unversioned entries.
* fix(router): skip misleading tags error when no candidates (e.g. cooldown)
Return early from get_deployments_for_tag when healthy_deployments is empty so
tag-based routing does not raise no_deployments_with_tag_routing after cooldown
filters all deployments. Adds regression test.
Made-with: Cursor
* feat(oci): add embedding support and update model catalog
- Add OCIEmbeddingConfig for OCI GenAI embedding models
- Add 16 new chat models (Cohere, Meta Llama, xAI Grok, Google Gemini)
- Add 8 embedding models (Cohere embed v3.0, v4.0)
- Update documentation with embedding examples
- Update pricing for all new models
* test(oci): add unit tests for OCI embedding support
- 17 unit tests covering OCIEmbeddingConfig
- Tests for URL generation, param mapping, request/response transform
- Tests for model pricing JSON completeness
* style(oci): format with black and ruff
* fix(oci): correct embedding request body format
OCI embedText API expects inputs, truncate, and inputType at the
top level of the request body, not nested under embedTextDetails.
Fixed transformation and updated tests accordingly.
Verified with real OCI API: 3/3 embedding models working.
* docs: clarify tag routing early return and test intent
Made-with: Cursor
* fix(oci): address code review findings from Greptile
- P1: Fix signing URL mismatch with custom api_base by accepting
api_base parameter in transform_embedding_request
- P2: Remove encoding_format from supported params (OCI does not
support it, was silently dropped)
- P2: Raise ValueError for token-array inputs instead of silently
converting to string representation
- Add test for token-list rejection
* fix(mcp): add STS AssumeRole support for MCP SigV4 authentication
MCPSigV4Auth only supported static AWS credentials or the boto3 default
credential chain. Production Kubernetes environments typically authenticate
via IAM role assumption (sts:AssumeRole), which was not possible.
Add aws_role_name and aws_session_name parameters to the MCP SigV4 auth
stack. When aws_role_name is provided, MCPSigV4Auth calls sts:AssumeRole
to obtain temporary credentials before signing requests. Explicit keys,
if also provided, are used as the source identity for the STS call;
otherwise ambient credentials (pod role, instance profile) are used.
* fix: stop logging credential values and add missing redaction patterns
Replaces raw credential values in debug/error log messages with
boolean presence checks or type names. Adds PEM block, GCP token,
JWT, SAS token, and service-account blob patterns to the redaction
filter. Fixes private_key pattern to capture full PEM blocks instead
of stopping at the first whitespace.
Addresses: Vertex AI credential JSON (including RSA private key)
being logged to stderr on health check failures.
* fix: log only field names for UserAPIKeyAuth, not full object
* style: apply black formatting to experimental_mcp_client/client.py
* style: fix black/isort formatting and mypy error in proxy_server.py
- Fix black formatting in experimental_mcp_client/client.py (done in prev commit)
- Fix black/isort formatting in key_management_endpoints.py, proxy_server.py, transformation.py
- Fix mypy: iterate over optional list safely (access_group_ids or []) in proxy_server.py
* fix(test): patch check_migration.verbose_logger directly to fix xdist ordering issue
When test_proxy_cli.py tests run before test_check_migration.py in the same
xdist worker, litellm.proxy.db.check_migration is already in sys.modules.
Patching litellm._logging.verbose_logger has no effect on the already-bound
reference. Patch the correct target (check_migration.verbose_logger) and
import the module before patching so the order doesn't matter.
* fix(mypy): make api_base Optional in PydanticAIProviderConfig to match base class signature
---------
Co-authored-by: Ihsan Soydemir <soydemir.ihsan@gmail.com>
Co-authored-by: Milan <milan@berri.ai>
Co-authored-by: Daniel Gandolfi <danielgandolfi@gmail.com>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: michelligabriele <gabriele.michelli@icloud.com>
Co-authored-by: user <70670632+stuxf@users.noreply.github.com>
Co-authored-by: Ishaan Jaffer <ishaanjaffer0324@gmail.com>
Tests cover: matching deployments, wrong team_id filtering, string-encoded
model_info, empty results, invalid model_info, and mixed deployment filtering.
Also updates MockPrismaClient.find_many to support the new startswith query.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat: add brave/search to model_prices_and_context_window.json (#25042)
Brave Search is supported by litellm as a search provider (documented at
docs.litellm.ai/docs/search/brave and listed in provider_endpoints_support.json)
but was missing from model_prices_and_context_window.json, making it invisible
to any code that discovers search providers from litellm.model_cost.
Cost: $0.005/query ($5 per 1,000 requests) per https://brave.com/search/api/
* feat(models): add NVIDIA Nemotron 3 Super 120B on Bedrock (#24588)
* feat(models): add NVIDIA Nemotron 3 Super 120B on Bedrock
Add model definition for nvidia.nemotron-3-super-120b-a12b-v1 via
Bedrock Converse API with pricing, context window (256k/32k), and
capability flags (function calling, tool choice, system messages).
* fix model ID to nvidia.nemotron-super-3-120b + add tests
Correct the Bedrock model ID from nvidia.nemotron-3-super-120b-a12b-v1
(NVIDIA's internal name) to nvidia.nemotron-super-3-120b (the actual
AWS Bedrock programmatic model ID). Add unit tests verifying model
resolution, pricing, and context window.
* fix(proxy): allow JWT auth for /v1/mcp/server sub-paths (#24698)
mcp_routes only contained "/v1/mcp/server" (exact match). Starlette's
compile_path produces an end-anchored regex, so sub-paths like
/register, /health, /submissions, /oauth/* all failed the JWT
allowed_routes_check. Add a {path:path} wildcard entry so all
sub-paths are covered.
---------
Co-authored-by: Daniel Yudelevich <4537920+yudelevi@users.noreply.github.com>
Co-authored-by: michelligabriele <gabriele.michelli@icloud.com>
Adds tenant_id, client_id, and client_secret to the Azure provider entry
in provider_create_fields.json so the credential add/edit modals and the
add-model form surface Service Principal auth as an alternative to api_key.
The Azure handler already reads these fields from litellm_params at request
time via get_azure_ad_token(); this change makes them inputtable from the
UI without code changes to the React components (the form is driven by
GET /public/providers/fields).
