38 commits

Author	SHA1	Message	Date
yuneng-jiang	6372ca32c1	Revert "chore(ci): sync litellm_internal_staging into daily OSS branch (#33337)" (#33339) ... This reverts commit 90f495f8dc.	2026-07-14 19:32:25 -07:00
yuneng-jiang	90f495f8dc	chore(ci): sync litellm_internal_staging into daily OSS branch (#33337) ... * feat(router): add LLM-based classifier option to complexity router (#32169) * feat(router): add LLM-based classifier option to complexity router Adds classifier_type: "heuristic" | "llm" to complexity_router_config. When set to "llm", the router calls a configured model (e.g. a small model like haiku) via structured output to pick the complexity tier, falling back to the existing regex/keyword scorer on any error, empty response, or unparseable output. * feat(ui): add classifier_type option to complexity router UI, fix edit flow Adds an "Advanced: Classification Method" section to ComplexityRouterConfig with a heuristic/LLM toggle, revealing a classifier model picker and timeout when LLM is selected. Also fixes the auto router edit modal, which never rendered the complexity router UI at all (it only handled the semantic router), and the "Edit Auto Router" button visibility check, which was gated on auto_router_config and never matched complexity router deployments. * fix(router): attribute classifier calls to caller, raise default timeout Forwards the original request's litellm_metadata into the classifier's acompletion call. Without it, the proxy's cost-tracking gate sees no user_api_key/team_id/user_id and silently drops spend logging and budget accounting for every classifier call, letting an authenticated user rack up unaccounted provider spend via repeated requests. Also raises the default classifier timeout from 400ms to 3000ms (400ms undershoots real LLM latency and would silently degrade to the heuristic scorer on most requests) and corrects the module/class docstrings, which still claimed zero external API calls after the llm classifier path was added. * fix(ci): resolve ruff strict-budget and frontend-lint failures - Use PEP 585 generics (dict/tuple/list) in the new aclassify/_classify_with_llm signatures instead of typing.Dict/Tuple/List, and suppress BLE001 on the intentionally broad except in aclassify's fallback path with a reason. - Fix prettier formatting in ComplexityRouterConfig.tsx. - Regenerate eslint-metrics.json (was stale after the classifier UI changes). * fix(ci): regenerate stale eslint-metrics.json * fix(router): strip parent budget reservation from classifier metadata The classifier's internal acompletion call previously forwarded the parent request's full litellm_metadata, including its budget reservation (user_api_key_budget_reservation / user_api_key_auth). That reservation belongs to the routed completion the classifier is deciding on, not to the classifier call itself, so it's now stripped while key/team attribution fields are still forwarded for spend logging. * fix(bedrock): add jp.anthropic.claude-opus-4-8 to model cost map (#32840) * fix(bedrock): add jp.anthropic.claude-opus-4-8 to model cost map * test: use apac regional profile for cost-map fallback test since jp now has an entry * fix(responses): preserve reasoning_tokens through chat->responses usage translation (#32837) * fix(responses): preserve reasoning_tokens through chat->responses usage translation Remove the unconditional else-branch that wrote reasoning_tokens=0 whenever completion_tokens_details.reasoning_tokens was None or absent. Also change OutputTokensDetails.reasoning_tokens from int=0 to Optional[int]=None so that re-instantiation without explicit reasoning_tokens no longer silently zeroes out the field, and remove the same hardcoded zero from the mock_responses_api_response initializer. * test(responses): update assertions to match Optional[int] reasoning_tokens default * fix(responses): preserve explicit reasoning_tokens=0 in usage translation Align the reasoning_tokens guard with the is-not-None guards used for text_tokens and image_tokens: a provider-reported zero passes through while an absent value stays omitted. --------- Co-authored-by: Deepanshu <deepanshu.lulla@alpha-sense.com> * fix(bedrock): gate in-place system role messages on model support for Claude Invoke (#32831) * fix(bedrock): gate in-place system role messages on model support for Claude Invoke * feat(bedrock): default unmapped Claude 4.8+ to in-place system role handling via fallback rule * fix(responses-api): raise APIError on in-stream error events; widen ErrorEventError.param to accept dict (#32835) * fix(responses-api): raise APIError on in-stream error events; widen ErrorEventError.param - BaseResponsesAPIStreamingIterator._maybe_raise_for_error_event inspects each chunk and raises litellm.APIError for type=error and type=response.failed events so callers see an exception instead of a benign stream chunk - rate_limit* codes map to 429; client error codes (invalid_request_error, context_length_exceeded, etc.) map to 400; all other codes default to 500; raw integer codes are never used as-is as HTTP status codes - ErrorEventError.param widened from Optional[str] to Optional[Union[str, Dict]] to prevent Pydantic ValidationError on dict-typed param payloads silently dropping error events before any type inspection * test(responses-api): add streaming iterator error event tests to CI-covered path * test(responses-api): cover response.failed, dict-error, null-error, and sync iterator paths * test(responses-api): set completion_start_time on mock logging objects for internal staging _process_chunk * fix(responses-api): map insufficient_quota to 429, derive failed-response log status from error code, and record failed-stream usage for spend accounting insufficient_quota moves out of the 400 bucket; OpenAI returns HTTP 429 for it and the non-streaming exception mapping treats 429 as RateLimitError, so the in-stream mapping now agrees _handle_logging_failed_response previously hardcoded APIError(status_code=500), so a rate-limited response.failed was logged to integrations as 500 while the caller saw 429; it now shares the same error-code-to-status mapping via _error_event_fields and _status_code_for_error_code usage carried on a response.failed event is now stashed as combined_usage_object with its computed cost on the logging object before failure handlers run, reusing the mid-stream-interruption spend recovery path (_failure_handler_helper_fn, proxy post_call_failure_hook, _ProxyDBLogger), so failed streams count their billed tokens instead of logging zero cost dedupe: TestMaybeRaiseForErrorEvent in tests/llm_responses_api_testing duplicated tests/test_litellm/responses/test_streaming_iterator_error_events.py, which is the canonical mirrored location and CI-covered via test-unit-responses-caching-types; the duplicate class is removed * fix(responses-api): wrap retriable in-stream errors in MidStreamFallbackError and map error type field to status Mirror chat streaming semantics from _handle_stream_fallback_error: 429 and 5xx in-stream error events now raise MidStreamFallbackError carrying the mapped APIError so the router's FallbackResponsesStreamWrapper triggers mid-stream fallback and cooldown; non-retriable 4xx still raise APIError directly. Status mapping now reads both the OpenAI error type and code fields, so type-classified client errors (e.g. invalid_request_error with code invalid_prompt) map to 400 instead of falling through to 500. * fix(responses-api): accumulate streamed output text so mid-stream fallback continues instead of restarting MidStreamFallbackError was always raised with generated_content="", so the router's stream_with_fallbacks treated every mid-stream error as pre-first-chunk and retried with the original input, streaming duplicated content to clients that had already received partial output. The iterators now accumulate response.output_text.delta text (mirroring chat's response_uptil_now) and pass it as generated_content, letting the router build a continuation input via _build_responses_continuation_input. * test(responses-api): pin in-stream token limit error to raised APIError --------- Co-authored-by: Deepanshu <deepanshu.lulla@alpha-sense.com> * fix(prometheus): skip budget metric DB lookups when gauges are NoOpMetric (#32834) adds a top-level guard in _increment_remaining_budget_metrics that returns early when all four budget gauges are NoOpMetric (excluded from prometheus_metrics_config), and per-entity guards in each _set_*_budget_metrics_after_api_request helper for partial disabling. eliminates four async DB/cache round-trips per successful LLM request when budget metrics are disabled. Co-authored-by: Deepanshu <deepanshu.lulla@alpha-sense.com> * fix(anthropic): strip @version suffix in _model_map_lookup_candidates (#32833) vertex_ai/claude-opus-4-8@default (and sibling @default models) were misclassified as non-adaptive because _model_map_lookup_candidates only stripped provider prefixes but never the @<suffix> portion. The lookup produced candidates like ["vertex_ai/claude-opus-4-8@default", "claude-opus-4-8@default"], neither of which exists in model_cost, so _is_adaptive_thinking_model returned False. LiteLLM then sent thinking.type=enabled to a @default Vertex AI endpoint that requires thinking.type=adaptive, resulting in a 400. _strip_version_suffix now removes @<suffix> from each candidate, adding the bare model name (e.g. "claude-opus-4-8") to the lookup chain. Also adds supports_adaptive_thinking: true to the three @default model_cost entries that were missing it as belt-and-suspenders. Co-authored-by: Deepanshu <deepanshu.lulla@alpha-sense.com> * fix(datadog): split log batches proactively under intake payload limits (#32860) * fix(datadog): split log batches proactively under intake payload limits * fix(datadog): size intake chunks with exact wire serialization * fix(anthropic): translate adaptive thinking/effort to pre-4.6 model support (#32867) * fix(anthropic): translate adaptive thinking/effort to pre-4.6 model support AnthropicMessagesConfig now reshapes the 4.6+ adaptive-thinking interface (thinking:{type:adaptive} + output_config:{effort:...}) to whatever the routed model supports. Thinking-capable non-adaptive models (e.g. Haiku 4.5, Sonnet 4.5) get the effort translated to a legacy thinking budget_tokens. Models with no reasoning support have thinking/effort dropped under drop_params. And because adaptive thinking carries no budget while the legacy form must satisfy Anthropic's max_tokens > budget_tokens rule, the translated budget is capped below max_tokens, dropping thinking when max_tokens can't fit the minimum budget. 4.6+ models pass through untouched. This matters because clients like Claude Code speak native Anthropic /v1/messages and send the adaptive interface unconditionally, regardless of the routed model. The native passthrough previously only capability-gated the OpenAI-style reasoning_effort alias and forwarded native output_config/adaptive thinking raw, so a pre-4.6 model rejected it with "This model does not support the effort parameter" and the request failed. Claude Code already gets drop_params auto-set, so its requests now succeed. * test(anthropic): gate undersized-max_tokens thinking drop on drop_params; add edge tests Addresses review feedback on the max_tokens-too-small branch. Previously a thinking-capable model whose max_tokens could not fit the minimum thinking budget had thinking silently dropped regardless of drop_params, while a residual output_config field in the same call still raised when drop_params was off. Gate both consistently on drop_params: raise a clear error (naming max_tokens for the undersized case) when drop_params is off, drop otherwise. Claude Code gets drop_params auto-set, so it still succeeds. Adds tests for the undersized-max_tokens raise, the residual output_config raise, and the no-adaptive-interface passthrough on a non-adaptive model. * fix(anthropic): make adaptive-effort translation silent to avoid breaking provider strip contracts The previous raise-when-not-drop_params behavior broke existing bedrock and vertex messages tests: those providers already silently strip unsupported output_config for pre-4.6 models (issue #22797) with no drop_params required, and the shared parent transform raising pre-empted that. It also conflicted with the goal of keeping requests working rather than failing them. Make the reshape silent: translate effort to legacy thinking for thinking-capable models, drop thinking for non-reasoning models, and remove only the consumed effort key from output_config, leaving any residual (e.g. format) for provider subclasses (bedrock/vertex) to handle. No raise, no drop_params gating. This also resolves the review note about inconsistent drop_params handling by making every path uniform. Updates the tests to assert the silent behavior and residual output_config preservation. * fix(anthropic): handle output_config-capable but non-adaptive models (Opus 4.5) Greptile caught a real bug: the early-return guard treated supports_output_config as equivalent to supporting adaptive thinking. Claude Opus 4.5 advertises supports_output_config (it accepts output_config.effort) but is not adaptive, so it rejects thinking:{type:adaptive} with "adaptive thinking is not supported on this model". The guard early-returned for Opus 4.5 and forwarded the adaptive thinking block raw, reproducing the exact failure the fix is meant to prevent. thinking:{type:adaptive} and output_config.effort are independent capabilities. Only early-return for adaptive-thinking models. For a model that supports output_config.effort but is not adaptive, keep the native effort and drop only the unsupported adaptive thinking block. Verified live against Opus 4.5: the Claude Code payload now returns 200 instead of 400. Adds regression tests for Opus 4.5 with and without adaptive thinking. * fix(anthropic): translate adaptive thinking for effort-capable pre-4.6 models Claude Opus 4.5 advertises supports_output_config but not adaptive thinking, so the early-return guard forwarded thinking.type=adaptive raw and Anthropic rejected it. The guard now only skips true adaptive models; effort-only requests on effort-capable models still pass through untouched. The _map_reasoning_effort call is wrapped to surface unrecognized effort values as a clean 400, matching _translate_reasoning_effort_to_anthropic * fix(anthropic): fall back to legacy thinking when effort level unsupported Opus 4.5 accepts output_config.effort but only low/medium/high; Claude Code defaults to xhigh on newer models, so preserving that level raw gets rejected by Anthropic. Gate the native-effort passthrough on _validate_effort_for_model and fall through to the budget translation for unsupported levels * fix(anthropic): keep effort-only requests untouched for provider normalization The xhigh fall-through consumed effort-only requests on effort-capable models, breaking bedrock invoke's own normalization which clamps xhigh to the model's ceiling after the base transform runs (test_bedrock_messages_normalizes_output_config_effort_for_opus). Restrict the fall-through to requests that carry adaptive thinking; effort-only requests pass through so provider subclasses keep owning level clamping --------- Co-authored-by: Abhimanyu Kapur <38531241+akapur99@users.noreply.github.com> * test(models): assert capability fields on regional Azure gpt-5.6 entries (#32875) * feat(auto_router): keyword tier overrides and semantic keyword matching for the complexity router Add deterministic keyword-to-tier overrides and optional embedding-based (semantic) keyword matching to the complexity router, and surface both in the Add Auto Router UI behind a Router Type selector: "Auto-Router v2 [Recommended]" (complexity tiers + keyword overrides + semantic matching, the default) and "Semantic Router [to be deprecated]" (the existing utterance-based router, unchanged). Keyword-to-tier overrides resolve to the highest tier matched rather than the first keyword matched, so match order no longer affects the routing decision. Backend: - config: KeywordTierRule model plus keyword_tier_rules, semantic_keyword_matching, embedding_model, and match_threshold on ComplexityRouterConfig, with a validator requiring an embedding model and rules when semantic matching is on - complexity_router: evaluate keyword rules before scoring; lexical matches escalate to the most-severe matched tier (order-independent), and semantic mode reuses LiteLLMRouterEncoder + SemanticRouter to match paraphrases by cosine similarity, falling back to the scorer when nothing matches - model management: clear complexity_routers on cache reload so config edits take effect Frontend: - Add Auto Router tab restores the Router Type radio (Auto-Router v2 recommended by default, Semantic Router still available) and sends keyword_tier_rules plus the semantic settings on the recommended path, instead of flattening keywords into custom_technical_keywords - client-side guard blocks submit when semantic matching is enabled without an embedding model or without any keyword tier rules, mirroring the backend validator - moved the "How Classification Works" explainer below Custom Technical Keywords and above Keyword Tier Overrides - remove the Test Connection action from the recommended flow, which can't build a valid pre-save payload for a router (leaves a TODO for a JSON preview / config test follow-up) Tests cover lexical escalation, semantic matching via the real library with injected embeddings, the semantic config guard, config validation, the reload-clear regression, and the frontend payload builder * fix(bedrock): flag mapped Claude 4.8+ entries with supports_mid_conversation_system (#32882) Exact cost-map hits resolve before fallback-generalization rules, so the mapped Sonnet 5, Fable 5 and jp Opus 4.8 Bedrock entries bypassed the bedrock-anthropic-claude-mid-conversation-system rule and hoisted mid-conversation system messages, invalidating the prompt cache. * feat(team): let org admins reach PATCH /team/{team_id} like POST /team/update Wire the coarse route gate so PATCH /team/{team_id} is reachable by exactly the roles that can call POST /team/update: proxy admins, org admins of the team's own organization, and JWT admins. Regular internal users and view-only proxy admins stay blocked, matching the existing endpoint Because the team id lives in the path rather than the body, the org-context resolver now also reads it from path_team_id for the bare /team/{team_id} route, so an org admin's organization is resolved and injected the same way it already is for POST /team/update. /team/{team_id} is added to management_routes rather than the role-agnostic self_managed_routes; the latter would have opened POST /team/new to any authenticated user through the shared /team/{team_id} path pattern * feat(ui): root the gateway breadcrumb in the AI Gateway selector The AI Gateway select (ViewSwitcher) now sits at the root of the DashboardHeader breadcrumb instead of on the right, so the top bar reads [AI Gateway select] > Page to match the redesign. It keeps the same dropdown, including the Chat / Chat UI options. When no plugins are registered and Chat UI is disabled there is nothing to switch between, so the breadcrumb falls back to the static section crumb rather than rendering a dangling leading separator * fix(complexity_router): build semantic route index once under concurrent cold-start Concurrent first requests each hit asyncio.to_thread to build the SemanticRouter index, firing duplicate embedding calls for the static route utterances. Guard the lazy build with a per-router asyncio.Lock (double-checked) so the index is constructed exactly once regardless of how many callers race in cold. Adds a regression test asserting ten simultaneous cold-start requests build the index the same number of times as a single request, and reworks the fake embedding router to count builds by how often a route utterance is embedded (robust to which embedding path the library uses) while still recording sync-call thread ids for the off-event-loop assertion. * feat(ui): always show the gateway selector with a discoverable Chat entry The AI Gateway selector now always renders at the breadcrumb root, even with no plugins and Chat UI disabled, so the Chat feature stays discoverable. The Chat entry is always listed: clickable when enabled, and disabled with an "Admins can enable in Settings" hint when it is off. Since the selector is now unconditional, the useViewSwitcherVisible hook and the section-crumb fallback added in the previous commit are removed * fix(proxy): guard delete_model router eviction on auto_router/ prefix delete_model popped the auto_routers/complexity_routers registries by the deleted deployment's model_name without checking it was actually an auto_router/* deployment. Deleting a regular DB model that merely shares a name with a config-defined router therefore evicted that router, which add_deployment never restores, leaving it unroutable until a proxy restart. This is the same cross-tenant DoS clear_cache was hardened against; mirror its auto_router/ prefix guard here. Extracts _deployment_name_and_model to read model_name and litellm_params.model from the deployment (delete_deployment returns the raw model_list dict at runtime despite its Deployment annotation), and adds a regression test asserting a same-named config router survives deletion of an unrelated regular model. * refactor(fallback-generalizations): split rules into routing and provider-neutral capability kinds * feat(fallback-generalizations): widen adaptive-thinking gate to any claude family at major 5+ * fix(fallback-generalizations): tolerate legacy remote rule schema and keep register_model cache-pricing inheritance * fix(fallback-generalizations): let exact cost-map entries beat capability rules across lookup-candidate ladders * fix(team): bound json merge patch recursion depth apply_json_merge_patch recurses into nested objects, which the repo's recursive_detector code-quality check flags because unbounded recursion over caller-supplied JSON has caused CPU/stack issues before. Cap the recursion at a depth far above any realistic team-metadata shape and reject deeper patches with a ValueError so a pathologically nested body fails closed instead of overflowing the stack, then register the function in the detector's ignore list alongside the other depth-bounded JSON walkers * fix(auth): tolerate request objects without path_params in common_checks The PATCH /team/{team_id} org-context wiring reads request.path_params to resolve the team id from the path. A real Starlette Request always exposes path_params, but common_checks is exercised with lightweight request doubles that don't, which raised AttributeError. Read it defensively so a missing or null path_params falls back to no path team id, matching the "not a bare team route" outcome; real requests are unaffected * fix(mcp): re-register DCR client when proxy origin no longer matches its registered redirect_uri A dynamically registered (RFC 7591) OAuth client persisted onto the MCP server row is bound to the redirect_uri it was first registered with, but that binding was never recorded. After the proxy's public origin changed, every authorize paired the reused client with the new callback and the IdP rejected it permanently. The DCR persist now records redirect_uris alongside the client identity. The admin register path treats a positive mismatch between the recording and the current callback as stale and re-registers a replacement client; rows without a recording (pre-existing installs and admin-configured clients) are grandfathered so upgrades never re-mint client_ids or orphan refresh tokens. The persist also writes client_secret and token_endpoint_auth_method explicitly as None when absent so the credential blob merge cannot pair a re-registered public client with the previous client's secret. Public register routes and non-admin callers keep existing behavior. Closes #32473 * fix(mcp): emit one operator warning per DCR re-registration event The stale-redirect path logged three warnings for a single re-registration: the staleness probe plus the reuse skip in both register_client_with_server and the persist race guard. The reuse-skip message is a mechanical consequence of the probe's decision, so it now logs at debug; the actionable warning that names both bindings and the re-authentication impact is emitted once by _persisted_dcr_redirect_uri_is_stale * ci: gate tests/e2e on zero basedpyright errors in pre-commit and lint CI * refactor(ui): use TanStack Pacer debounce for the team keys search Replace lodash/debounce in TeamVirtualKeysTable with useDebouncedValue from @tanstack/react-pacer, matching the sibling VirtualKeysTable and PaginatedKeyAliasSelect which already debounce their key-alias search that way. Pacer is already a dependency, so this drops the odd-one-out lodash usage and keeps the search-debounce pattern consistent across the key tables. * fix(fallback-generalizations): cover bare Claude majors in baseline and routing, require claude- prefix in adaptive gate * fix(mcp): strip scheme default port from get_request_base_url netloc * feat(ui): typed openapi-fetch foundation (fetchClient) + first typed caller (useCustomers) (#29884) * feat(ui): add the typed openapi-fetch client (fetchClient) as the dashboard fetch foundation Introduces fetchClient (openapi-fetch) bound to schema.d.ts, used inside ordinary TanStack Query hooks so path/query/body types come from the proxy's OpenAPI spec. A small runtime registry feeds the client the base URL and auth header name (registered by networking) and the session token (published by AuthContext), so call sites carry no token plumbing; auth-header injection and ApiError mapping live in openapi-fetch middleware reusing deriveErrorMessage/ApiError from client.ts, and non-2xx maps to a thrown ApiError so query functions just read .data. The base URL default resolves from NEXT_PUBLIC_BASE_URL so a request still targets the right origin if it fires before networking registers its getter. AuthContext clears accessToken alongside the token on logout so no query fires unauthenticated after the session ends. Foundation only; callers migrate one at a time, each fully typed, in follow-up changes. * feat(ui): migrate useCustomers to the typed fetchClient Converts useCustomers from allEndUsersCall to fetchClient.GET("/customer/list"); the response is typed as LiteLLM_EndUserTable[] from the schema, so the hand-written Customer/CustomersResponse types are deleted. They were also inaccurate (allowed_model_region was string but is "eu"|"us", and a budget_id the table has no field for). No cast; the schema type flows to the one consumer. First caller on the new pattern. * fix(ui): route typed-client errors through the session-expiry handler The typed fetchClient middleware threw ApiError without invoking the handleError side effect that the legacy createApiClient wires via onError, so a migrated caller hitting an expired key no longer triggered the auto-logout. Add an error-handler seam to runtime.ts, register handleError from networking.tsx alongside the base-url/header getters, and call it in the middleware before throwing so both clients behave the same. Regression test asserts the handler fires with the derived message on non-2xx and stays silent on success * fix(ui): point the customers EndUser type at CustomerResponse The /customer/list response model was renamed to CustomerResponse on staging; the merged branch still aliased EndUser to LiteLLM_EndUserTable, so the exported type and its test mock had drifted from what the schema actually returns. CustomerResponse is also the accurate shape (it types allowed_model_region as 'eu' | 'us' and carries budget_id) * chore(ui): refresh eslint-metrics baseline after staging merge The recorded baseline predated the litellm_internal_staging merge, so its no-explicit-any and no-large-inline-object-arg counts were higher than the merged tree actually has. Regenerate via npm run lint:metrics so the gate reflects current reality * refactor(ui): source the typed client token from the session cookie, not AuthContext The typed client read its bearer from a runtime value that AuthContext pushed via setAuthToken, but migrated hooks gate enabled on useAuthorized, which decodes the cookie directly. Two independent derivations of the same cookie with different timing: on first load the query fires (useAuthorized sees the token) before AuthContext's async effect publishes it, so the first request goes out unauthenticated and only succeeds on a React Query retry. Make the token a registered getter like the base-url and header-name getters, reading the same cookie useAuthorized decodes, so the client's token and the gate can't diverge. Revert the AuthContext changes entirely; nothing is pushed from React state anymore. * test(e2e): cover Langfuse logging.yaml P0 logs_spend cells (#32857) * test(e2e): cover Langfuse logging.yaml P0 logs_spend cells Team, user/key, and org-scoped dynamic Langfuse callbacks drive real chat traffic and assert calculatedTotalCost matches StandardLogging response_cost and proxy spend. Also assert tool calls and applied guardrails land on the trace. Missing env or proxy is a hard failure, never a skip * test(e2e): use langfuse_otel callback for Langfuse spend coverage Team and key dynamic logging attach callback_name=langfuse_otel (OTLP to Langfuse) instead of the classic langfuse SDK. Match generations named litellm_request by prompt marker and user_api_key_alias * test(e2e): require Langfuse spend assert; drop AGENTS.md Guardrail path no longer soft-gates logs_spend. Non-stream responses must return positive x-litellm-response-cost; remove tests/e2e/AGENTS.md * test(e2e): fail when Langfuse spend is missing on guardrail path Always run logs_spend assertions for tool_permission; require positive x-litellm-response-cost on non-stream and positive /spend/logs spend * test(e2e): do not fall back to unmatched spend log rows poll_proxy_spend_for_key returns None when response_id or positive-spend filters match nothing, instead of silently using rows[0] * fix(complexity_router): use max aggregation for semantic keyword route scoring SemanticRouter defaults to mean aggregation across a route's utterances. Since each tier's route holds one utterance per configured keyword, a real semantic match on one keyword was averaged together with the tier's other, unrelated keywords and dragged below match_threshold — e.g. a MEDIUM tier with keywords [beep, boop, new york] never fired for a genuine "new york" paraphrase, because mean(sim_to_beep, sim_to_boop, sim_to_new_york) landed well under the threshold even though sim_to_new_york alone cleared it. Pass aggregation="max" so a tier matches if the query is close enough to any one of its keywords, not the average of all of them. Verified against live Voyage embeddings: raw cosine similarity for "new york" vs a paraphrase was 0.54 (above a 0.5 threshold), but the route scored 0.28 under mean aggregation and never matched; max aggregation fixes it. Adds a regression test with a tier holding one matching and two unrelated keywords, asserting the tier still fires; fails without aggregation="max". * refactor(auth): resolve PATCH team org-context from the route template Replace the request.path_params read (and its defensive getattr guard) with the route template. A real Starlette request always exposes path_params, but common_checks runs on lightweight request doubles that don't, so reading it directly forced a getattr workaround that only existed to tolerate those doubles. Instead, match the route template (/team/{team_id}) to identify the RESTful update route and take the team id from the last path segment. This drops the path_params dependency entirely, and because the template distinguishes the PATCH route from its single-segment siblings (/team/new, /team/list, ...), it also avoids a spurious team lookup those routes would otherwise trigger if we matched the resolved path shape alone. * chore(ui): remove eslint-metrics.json lint-count snapshot The eslint-metrics.json snapshot duplicated the violation counts already enforced by eslint-budgets.json. Keeping it current added a CI drift check, a pre-commit regenerate-and-flag step, and a standalone npm run lint:metrics script, none of which caught anything the budget gate did not, yet all of which failed noisily whenever the snapshot went stale. This drops the file and that machinery while leaving eslint-budgets.json as the actual ratchet gate * fix(complexity_router): preserve user_api_key_auth in sub-call metadata Removing user_api_key_auth entirely from classifier/embedding sub-call metadata (as _BUDGET_RESERVATION_METADATA_KEYS previously did) prevented _filter_deployments_by_model_access_groups from scoping those sub-calls to the caller's authorized access groups. An access-group-scoped caller could therefore reach embedding/classifier deployments outside their group. Only strip user_api_key_budget_reservation, which is the actual budget- reservation state that must not reach sub-calls. user_api_key_auth is now kept so access-group filtering works correctly for both the embedding path and the LLM classifier path. * test(e2e): drop vertex from pipecat tool smoke (#32925) Exclude vertex_ai from pipecat tool smoke; raw-ws tool_call_round_trip remains the Vertex source of truth. Also remove the Playwright key models dropdown suite so stage is not blocked by that UI harness * fix(complexity_router): sanitize budget reservation inside forwarded user_api_key_auth * fix(complexity_router): review hardening - blank keywords, router registry eviction, edit-modal controls - config: KeywordTierRule now strips and drops blank/whitespace keywords (a stray "" makes _keyword_matches match every prompt, silently forcing that tier for all traffic); still requires at least one real keyword to remain - frontend build_complexity_router_config: trim keywords and drop rules left empty so an unfilled "Add keyword rule" row no longer ships a rule the backend rejects with a 400 in the heuristic (non-semantic) flow, where the client-side semantic guard doesn't run - proxy clear_cache / delete_model: the auto_router/ prefix also covers quality_router/ and adaptive_router/, so pop the model_name from all four router registries (no-op where absent) instead of only auto/complexity; otherwise a DB quality_router's stale entry made reload raise "already exists" and abort, and adaptive left a leak - frontend ComplexityRouterConfig: only render the Keyword Tier Overrides and Semantic keyword matching sections when their change handlers are provided, so the edit-auto- router modal (which omits them) no longer shows interactive-but-dead controls * fix(anthropic): thread real provider through capability probes instead of pinning anthropic * docs(anthropic): note the two provider params' roles in _map_reasoning_effort * fix(anthropic): override custom_llm_provider in provider config subclasses so capability probes use the right namespace * feat(mcp): admit dcr_bridge oauth_delegate clients via a single envelope bearer * fix(mcp): admit dcr_bridge envelopes under the live key, not a frozen identity The bridge envelope sealed only user_id/server_id, and admission fabricated a UserAPIKeyAuth(user_id=...) with no object_permission, team_id, org_id, or key identity. Downstream MCP permission checks read the missing restrictions as unrestricted, so a caller holding a valid envelope for a restricted key could reach tools and servers that key was never granted, and a revoked key kept working until the envelope expired. Bind the hashed authorizing key into the envelope identity and reload the live UserAPIKeyAuth by it at admission via get_key_object, failing closed with a 401 when the key is missing, blocked, or expired. Authorization is resolved fresh per request instead of frozen at mint time, so current key/team/org and tool restrictions plus revocation are enforced. * fix(mcp): enforce team block and alias-priority token injection on bridge admission Two follow-ups on the envelope admission arm flagged in review. Team revocation bypass: _reload_admitted_key checked only the key's own blocked/expires, so blocking a key's team left every envelope minted under it live until expiry. Reload the team and reject a blocked team, mirroring common_checks, so a team block revokes its envelopes immediately. Caller-overridable upstream token: egress resolves the per-server auth header alias-first, but injection keyed under server_name, so for a server with a distinct alias a caller-forwarded x-mcp-{alias}-authorization sat at the higher-priority slot and paired the admitted identity with an attacker's upstream credential. Inject under alias-first so the sealed token owns the slot egress resolves. * fix(mcp): route bridge admission through the centralized policy gate and mirror the SCIM owner check * fix(mcp): import assert_never from typing_extensions for Python 3.10 * fix(mcp): surface real status from bridge admission policy gate instead of flattening to 401 Over-budget rendered 401 (should be 429), model-access and other typed failures collapsed to 401, and a transient DB outage was masked as an auth error. Mirror UserAPIKeyAuthExceptionHandler: budget maps to 429, a sub-check's own HTTPException/ProxyException keeps its status, a DB outage is a retryable 503, and only a genuinely unresolvable failure stays the fail-closed 401. * fix(rate-limit-v3): populate x-ratelimit-* remaining/limit values in standard_logging_object for streaming (LIT-4333) (#32711) Streaming requests return from common_request_processing before async_post_call_success_hook runs, so response._hidden_params.additional_headers never gets the v3 x-ratelimit-{descriptor_key}-{remaining|limit}-{rate_limit_type} entries. Prometheus / logging callbacks that read those values from standard_logging_object.hidden_params.additional_headers then see nothing; combined with the pre-existing gap that Prometheus reads from that same slot (LIT-2577 / PR #28816), per-key remaining RPM/TPM cannot be monitored for streaming traffic at all. Fix in three parts: - Stash the pre-call RateLimitResponse in the metadata channels the async success-logging callback inherits, alongside the existing top-level entry the non-streaming path reads. - Add async_logging_hook to the v3 handler. It fires in a distinct earlier loop inside async_success_handler (all callbacks' async_logging_hook complete before any async_log_success_event starts), so mirroring the pre-call snapshot into standard_logging_object.hidden_params.additional_headers and response._hidden_params.additional_headers here guarantees every downstream success callback sees the values regardless of registration order. Non-streaming keeps the existing async_post_call_success_hook write and this hook re-populates the same values idempotently. - Extract the shared `_merge_ratelimit_statuses_into_additional_headers` helper the non-streaming path already had inlined so both callsites emit the identical key shape. * fix(proxy): skip None model_name in clear_cache router eviction set * fix(mcp): map a DB outage during bridge key reload to a retryable 503 get_key_object's raw transport error propagated uncaught out of _reload_admitted_key as an opaque 500; classify it via the shared _raise_503_if_db_unavailable helper (also used by the live-policy gate) so a database outage is a retryable 503, while a key-not-found ProxyException stays the fail-closed 401. * test: remove live OpenAI fine-tuning job-creation test blocked by platform wind-down (#32933) OpenAI is winding down self-serve fine-tuning and the org can no longer create fine-tuning jobs (403 training_not_available; the CI key surfaces it as a 500 server_error), so test_create_fine_tune_jobs_async fails on every batches_testing run since 2026-07-11 and reruns never clear it. The request contract stays covered by the mocked create/list/cancel/ retrieve tests in the same file, and the deleted test's unique standard_logging_object assertions now run inside test_mock_openai_create_fine_tune_job. * refactor(anthropic): consolidate the provider fallback into a _resolved_provider property * feat: add lite auth print-token for Claude Code apiKeyHelper support (#32846) * feat: add silent CLI token refresh for apiKeyHelper support lite auth print-token prints a valid proxy credential for use as Claude Code's apiKeyHelper, transparently refreshing it first if the cached JWT is stale. This unblocks MDM-managed apiKeyHelper deployments (managed via `lite auth print-token`) that need silent mid-session credential rotation without restarting the client. Refresh capability is backed by a virtual key minted with an empty model list and cli_refresh metadata, kept strictly separate from the actual (short-lived, real-model-scoped) call credential -- so a leak of the credential that flows through every LLM request and subprocess env var can't also self-renew. The refresh flow is single-use: /sso/cli/refresh mints a fresh JWT + refresh token pair and blocks the presented refresh token immediately, so a replay can't mint a second pair from it. Server: /sso/cli/refresh (rotate) and /sso/cli/logout (revoke) endpoints. lite login now also stores a refresh token; lite logout revokes it server-side instead of only clearing the local file. * fix: allow non-admin users to hit CLI refresh routes; resolve apiKeyHelper base_url from token.json Found via a live end-to-end test against a real proxy + real Claude Code session: /sso/cli/refresh and /sso/cli/logout were unreachable for any non-proxy-admin caller, since Depends(user_api_key_auth) pulls in a route-RBAC gate that 403s any route not on an explicit allowlist. That made the feature unusable for actual end users, who authenticate as internal_user. Add both routes to internal_user_routes; the handlers already do their own fine-grained check (metadata.cli_refresh) same as /key/block does today. Also: `lite auth print-token` required an explicit --base-url/ LITELLM_PROXY_URL matching the stored token's origin, defaulting to localhost:4000 otherwise. But apiKeyHelper is configured bare (no flags), so this always mismatched a real deployment. Track whether --base-url was explicitly passed (via click's ParameterSource) and, if not, resolve the server from token.json directly instead of the CLI default. * test: mock refresh-token minting in test_cli_poll_key_tolerates_missing_user_row Landed on litellm_internal_staging after this branch's refresh-key minting change; needs the same mock as the other cli_poll_key tests since minting now runs unconditionally whenever a JWT is generated. * fix(ci): update test_cli_auth.py for refresh_token contract, regenerate schema.d.ts _poll_for_authentication now always includes "refresh_token" in its returned dict, and _handle_team_selection_during_polling returns a dict instead of a bare JWT string -- test_cli_auth.py predates this branch's refresh-token work and still asserted the old shapes. schema.d.ts regenerated via `npm run gen:api` to pick up the new /sso/cli/refresh and /sso/cli/logout routes (plus unrelated drift from other PRs merged since it was last generated). * fix(ci): apply CI's own schema.d.ts diff (enterprise routes I can't generate locally) Local `npm run gen:api` only sees OSS routes -- this machine's litellm_enterprise editable install points at a now-deleted temp directory, so it silently drops enterprise-only routes from the spec. Applied the exact diff CI's own generation produced instead of re-running the generator locally. * fix: close refresh-token race, fail closed on DB down, fix logout base_url Addresses Greptile review findings on the CLI refresh-token PR: - cli_refresh_token minted a new JWT + refresh token BEFORE blocking the presented one. Two concurrent requests bearing the same refresh token could both pass auth and both mint fresh pairs, yielding four live credentials from one consumed token. Now the presented token is consumed atomically first via update_many (only succeeding if it flips blocked from False/None to True); the loser gets count=0 and is rejected before anything is minted. - When prisma_client is None, refresh silently returned a new JWT without ever being able to mark the presented token consumed, leaving it valid indefinitely. Now fails closed with a 500 instead. - `lite logout` sent its revocation POST to ctx.obj["base_url"], which defaults to localhost:4000 when --base-url isn't passed -- the same bug print_token had before the base_url_explicit fix, just missed here. Now resolves the same way: trust the stored token's origin unless the caller explicitly overrode --base-url. * fix(ci): satisfy ruff format and narrow token_data type in logout * fix(security): never trust refresh-token metadata for authorization Addresses a real privilege-escalation path Veria flagged: cli_refresh_token read team_id, team_alias, and max_budget straight off the presented token's own metadata and used them to authorize the new JWT. Since any authenticated user can self-mint a virtual key with arbitrary metadata via the ordinary /key/generate endpoint, a self-forged key with {"cli_refresh": true, "team_id": "<any-team>", "max_budget": 999999999} would sail through _require_cli_refresh_token's only check (metadata.cli_refresh == True) and get a JWT scoped to a team the caller never belonged to, with a budget it never had -- full cross-team / budget bypass, and a removed team member could keep refreshing team-scoped sessions indefinitely. Metadata's team_id is now treated as an untrusted UX hint only: honored solely if the CALLER (identified by the authenticated key's own user_id, not client input) is a current member per a fresh get_user_object lookup. team_alias and max_budget are never read back from metadata at all -- team_alias comes from a live get_team_object lookup and max_budget is recomputed with the exact same capping logic the initial SSO login poll uses. _mint_cli_refresh_token no longer accepts or stores team_alias/max_budget, only the team_id hint. Added regression tests proving: a forged/stale team_id is dropped (falls back to no team, not silently honored), and a forged max_budget in metadata never reaches the issued JWT. * fix(ci): catch HTTPException specifically instead of bare Exception (BLE001) * fix: un-consume refresh token if minting the replacement fails Greptile flagged a real reliability gap: cli_refresh_token blocks the presented token atomically, then does several more DB calls before returning a replacement (user lookup, team lookup, JWT mint, new refresh-key mint). Since this endpoint exists specifically for fully unattended apiKeyHelper operation, a single transient failure in that window (DB hiccup, etc.) permanently stranded the user: their old token was already dead and no new one was issued, with no recovery path short of a full interactive browser re-login. Wrap that window in try/except; on any failure, best-effort revert the consumed token back to usable (blocked=False) before re-raising, so a retry can succeed. Standard compensating-action pattern since generate_key_helper_fn doesn't take an injectable transaction, so wrapping the whole thing in a real DB transaction isn't practical here. * fix(security): refresh key had unrestricted model access, not none Critical bug: _mint_cli_refresh_token used models=[] intending "no LLM access", but that's backwards in this codebase. Per _check_model_access_helper: `len(filtered_models) == 0 and len(models) == 0` -> all_model_access = True. An empty models list on a key with no team_id means UNRESTRICTED access to every model, not zero access. The CLI refresh token -- meant to be usable for nothing but silently exchanging itself for a new JWT -- was actually a fully unrestricted API key for its entire 90-day lifetime, completely undermining the whole point of keeping it separate from the short-lived call credential. Fixed with two independent layers: allowed_routes hard-restricts the key to exactly /sso/cli/refresh and /sso/cli/logout (the real enforced boundary, checked in the shared user_api_key_auth dependency for every route); models is set to an unmatchable sentinel string as defense-in-depth in case any code path only consults the models field. Added an end-to-end regression test that exercises the actual model-access-control function against a key shaped like the minted refresh token, rather than only asserting on what arguments were passed to the key-generation call -- the latter kind of test is exactly what let the original bug ship, since asserting `models == []` is equally consistent with "no access" and "unrestricted access" without checking what the access-control code actually does with that shape. Also: the compensating-rollback added for reliability un-blocked a consumed refresh token even when the underlying user no longer exists. That's a permanent, intentional rejection, not a transient failure -- un-blocking it would let a stale refresh token become valid again for a different account if the user_id is ever reused/re-registered. Moved the user-existence check outside the rollback-on-failure block so it stays permanently blocked. * refactor: rotate CLI refresh tokens via regenerate_key_fn instead of hand-rolled consume/rollback The refresh token is already a plain litellm virtual key, so rotation can delegate to the same atomic DB update /key/regenerate uses instead of a bespoke update_many + compensating-rollback dance. This makes silent CLI refresh an Enterprise feature, same as regular key regeneration. * refactor: replace CLI stateless JWT + refresh-key pair with one self-rotating virtual key The CLI previously minted two credentials on login: a stateless self-signed JWT for LLM calls, and a separate DB-backed refresh-only key (scoped away from ever calling an LLM) just to authorize minting a new JWT. Collapse this into a single real virtual key, used directly as the LLM bearer token and re-presented to /sso/cli/refresh to rotate its own secret in place. This also means the CLI session key now shows up in the Admin UI's Keys page and can be revoked/regenerated like any other key, rather than being an invisible, unmanageable stateless token. * refactor: drop silent CLI refresh, key just expires and requires re-login /sso/cli/refresh only ever benefited Enterprise deployments (regenerate_key_fn's gate), while everyone else already fell through to "re-run lite login" on failure. Cut the endpoint, the rotation logic, and the client-side refresh path entirely; print-token now just prints the cached key until it hits its LITELLM_CLI_JWT_EXPIRATION_HOURS duration, then fails fast telling the user to log in again. Session key itself is unaffected: still a real, revocable virtual key visible in the Keys UI, `lite logout` still revokes it directly. * fix(ci): regenerate schema.d.ts after removing /sso/cli/refresh route * revert: go back to stateless JWT, keep only lite auth print-token The virtual-key redesign (revocable, Keys-UI-visible credential) wasn't needed just to support print-token, and cost real server-side surface (a mint path, a logout-revoke endpoint, migrated tests/docs) for a property this repo doesn't need yet. Reverting cli_poll_key/_types.py/schema.d.ts back to the original stateless-JWT design; the only durable addition from this whole effort is `lite auth print-token` (reads the cached credential, prints it while fresh, fails with a clear message once it's past LITELLM_CLI_JWT_EXPIRATION_HOURS) plus the base_url_explicit plumbing it needs. `lite logout` goes back to clearing the local file only, since a stateless JWT can't be revoked server-side. * refactor: move CLI token freshness check to cli_token_utils, drop unnecessary renames Addresses review: the freshness check is a pure token-shape/timestamp util, not command logic, so it belongs alongside the other SDK-level CLI token helpers (load_cli_token, get_litellm_gateway_api_key) rather than in commands/auth.py. Also reverted a few incidental jwt_token/ session_key variable and string renames that weren't load-bearing. * fix(mcp): run the route gate on bridge admission so allowed_routes are enforced The envelope arm reloaded the identity and ran _run_centralized_common_checks but skipped RouteChecks.should_call_route, which the standard pipeline runs between the builder and common_checks. Because the centralized checks treat MCP as an inference route and never re-check allowed_routes, a key barred from MCP routes could mint an envelope at the token endpoint (not itself an MCP route) and replay it against MCP. Run the route gate before admitting, and clear the request-scoped budget_reservation, matching the wrapper's sequence; a disallowed route now surfaces the gate's own 403. * fix(proxy): reserve budget for tiered pricing Ensure tier-only models reserve their estimated request cost so concurrent requests cannot bypass exhausted budgets. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(proxy): bill tier-only deployments instead of $0 Route cost calculation to the deployment's router_model_id entry when it carries tiered_pricing but no flat per-token rate, so models like dashscope/qwen3.7-plus are billed via their tier table rather than the pricing-stripped shared alias. Co-authored-by: Cursor <cursoragent@cursor.com> * refactor(ui): convert activity metrics charts to shadcn/recharts (#32726) * refactor(ui): convert activity metrics charts to shadcn/recharts Swap the seven tremor AreaChart/BarChart sites in activity_metrics.tsx to the shared shadcn/recharts wrappers and switch CustomLegend/CustomTooltip to the ported versions in shared/charts. Chart props, colors, formatters, and legend behavior are unchanged; tests now assert on real recharts SVG output instead of tremor mocks. * fix(ui): restore tremor No data placeholder for empty AreaChart data * test(ui): scope activity metrics chart assertions to card titles instead of render order * feat(ui): extend topnav border across the sidebar header (#32920) Pin the sidebar header to the same 56px height as the dashboard topnav and give it a matching bottom border, so the two borders sit flush and read as one continuous line. Revert to auto height when the rail is collapsed so the stacked logo and toggle are not clipped. * fix(cost): coerce string tiered-pricing costs and share tier helper YAML-parsed tier costs can arrive as strings (e.g. "4e-07"), which broke arithmetic in the graduated tiered-pricing calculation. Coerce per-token costs to float in both the in-range and remaining-tokens paths. Move the tiered-cost helper out of the Dashscope module into a provider-neutral home so the proxy budget reservation no longer depends on a provider-specific module. Co-authored-by: Cursor <cursoragent@cursor.com> * docs(anthropic): clarify the Opus 4.5 branch in adaptive-effort translation Add an inline comment explaining that the effort-capable non-adaptive branch in _translate_adaptive_effort_for_non_adaptive_model exists for models like Claude Opus 4.5 that accept output_config.effort but reject adaptive thinking, and why effort-only requests pass through while adaptive requests with an unsupported effort level fall through to the legacy translation. * fix(anthropic): translate raw adaptive thinking for chat completions on pre-4.6 models Clients that pass thinking={"type": "adaptive"} directly (not via the reasoning_effort alias) on the /chat/completions interface had it forwarded unmodified to pre-4.6 Anthropic models, which reject the shape. Mirrors the translation already applied on the native /v1/messages passthrough (#32867): translate to legacy thinking={type: enabled, budget_tokens}, capped below max_tokens, dropping thinking when max_tokens can't fit even the minimum budget. Hoists the shared budget-capping helper onto AnthropicConfig so both paths use one implementation. * fix(proxy): reserve tiered budget all-or-nothing across all deployments Alibaba Model Studio (Dashscope) tiered pricing is all-or-nothing: the tier is selected by a request's total input tokens and every token, input and output, is billed at that one tier's rate. The reservation path used graduated slicing and, worse, picked the output tier from the output-token count, so a long-context request with a large output allowance reserved far less than the provider charges and could slip past a depleted budget. Select the tier from input tokens and apply its rates to all input and output tokens. Reservation also read tiered pricing from only the first deployment in a model group. A caller could hit an alias whose cheaper deployment was listed first and exceed the budget once routed to a costlier sibling. Estimate against every eligible deployment's pricing and reserve the maximum. Co-authored-by: Cursor <cursoragent@cursor.com> * feat(complexity_router): log the cause of each routing decision The complexity router's info log didn't say what drove a routing decision. Literal and semantic keyword matches logged an identical "keyword rule fired" line (no way to tell which mechanism fired), and the scorer's line carried no consistent marker tying it to the same question. Emit one greppable line per decision naming the cause: literal_keyword_match, semantic_keyword_match, or complexity_scorer. The hook already knows which ran (the config's semantic_keyword_matching flag distinguishes lexical from semantic; the override-vs-scorer branch distinguishes keyword match from scorer), so this is label-only: no behavior change, no new types, no added latency. Adds regression tests asserting each decision path logs its cause; they fail if a label is swapped or the cause= marker is dropped. * feat(ui): add redesigned sidebar account menu (#32931) * feat(ui): add redesigned sidebar account menu Introduce SidebarAccountMenu, a sidebar-only account/logout menu built on shadcn Popover/Switch/Badge/Separator/Button, and wire it into leftnav in place of the shared UserDropdown. The panel has a LiteLLM header with the bouncing moon and a clickable version tag, Tier/Role/Email/User ID rows with copy actions, the five display toggles, and Logout. UserDropdown is left untouched so the control-plane / chat navbar keeps its existing menu. The version tag links to the same release notes page as the navbar tag, and the bouncing icon reuses the existing header animation gated by the Hide Bouncing Icon toggle. * test(ui): point account-menu e2e specs at the migrated sidebar menu The sidebar account menu moved from an antd Dropdown to a Base UI popover (SidebarAccountMenu), so the login, logout, proxy-logout-url, and internal user identity specs were still waiting on antd-era locators (.ant-dropdown, the popupRender wrapper class, the user-dropdown-panel test id, and a menuitem-role Logout). Point them at the new panel test id (sidebar-account-menu-panel) and the button-role Logout instead. The logout behavior is unchanged since both menus call the same useLogout handler. * fix(bedrock-converse): translate adaptive thinking for pre-4.6 models Follow-up to #32867 (native /v1/messages) and the /chat/completions commit earlier on this branch, extending the same adaptive-thinking translation to the Bedrock Converse path. Clients like Claude Code send thinking={type: "adaptive"} on every request. When routed via Bedrock Converse to pre-4.6 models (claude-haiku-4-5, claude-sonnet-4-5), this was forwarded as-is and rejected by the model. Mirrors the translation already applied on the /chat/completions and /v1/messages paths: map to legacy thinking={type: enabled, budget_tokens}, capped below max_tokens. Also fixes the missing custom_llm_provider arg in the chat completions path's call to AnthropicConfig._map_reasoning_effort. * fix(mcp): run proxy-wide pre-DB gates on bridge envelope admission The envelope arm bypasses user_api_key_auth, so it never ran pre_db_read_auth_checks (request-size and body-safety limits, the IP allowlist, and the general_settings route allowlist) that the normal MCP admission path runs before any key lookup. A caller blocked by IP or a disallowed proxy route could be admitted through an envelope where the same principal on the normal path is rejected. Run those gates before the envelope crypto, mirroring the pipeline's pre-DB ordering; a blocked IP or route surfaces its own 403. * fix(anthropic): pass resolved provider to adaptive-thinking check The rebase onto staging changed _is_adaptive_thinking_model to require custom_llm_provider (no default), so the one-arg call in the raw adaptive thinking branch raised TypeError at runtime for any /chat/completions caller sending thinking={type: adaptive}. Use self._resolved_provider, matching the reasoning_effort branch just below. Caught by Greptile. * test(bedrock-converse): cover adaptive-thinking drop when max_tokens too small Adds the regression test for the warning-drop branch in the Converse adaptive-thinking translation, mirroring the chat completions path's test_raw_adaptive_thinking_dropped_when_max_tokens_too_small. * fix(guardrails): filter Add-Guardrail mode dropdown per provider (#32712) * fix(guardrails): filter Add-Guardrail mode dropdown per provider The GET /guardrails/ui/add_guardrail_settings endpoint returned every GuardrailEventHooks value in one flat supported_modes list, so the Admin UI rendered pre_mcp_call as a selectable Mode for every guardrail. Saving Content Filter or Tool Permission with pre_mcp_call then failed with a 400 because those guardrails' server-side supported_event_hooks list excludes it. Expose each guardrail's supported hooks as a get_supported_event_hooks classmethod on CustomGuardrail (mirrors the existing get_config_model pattern) and have the endpoint iterate guardrail_class_registry to build a supported_modes_by_provider map. The UI Mode dropdown filters by that map when the selected provider is known and falls back to the global list otherwise. __init__ now sources its own supported_event_hooks list from the classmethod so the two sides can't drift. Also register BedrockGuardrail, ToolPermissionGuardrail, lakera, lakera_v2, and presidio in guardrail_class_registry so they participate in the map (they were previously only in guardrail_initializer_registry and had no class-registry entry). Behavior change: guardrails that previously had no supported_event_hooks declared (aim, javelin, azure/text_moderation, cato_networks, crowdstrike_aidr, headroom, hiddenlayer, lasso, noma, onyx, prompt_security, qualifire, repelloai, zscaler_ai_guard, aporia_ai, lakera_ai, lakera_ai_v2, mcp_jwt_signer, model_armor, presidio) now validate the configured mode at instantiation. Existing configs where the mode was silently a no-op will fail at proxy startup with a clear validation error rather than running as a broken guardrail. Resolves LIT-4226 * fix(guardrails): add LITELLM_STRICT_GUARDRAIL_MODES escape hatch, preserve current mode in edit form Address Greptile P1 (startup break) and P2 (edit form UX): LITELLM_STRICT_GUARDRAIL_MODES defaults to true (raise on unsupported event_hook, unchanged behavior for the guardrails validated pre-PR). Setting it to false logs a warning and continues, giving deployments an opt-out while they fix configs that now surface as errors instead of silently no-op'ing. Regression test covers both modes. Edit form now surfaces the currently-saved mode even when it is not in the filtered per-provider list, so a legacy row (e.g. content_filter saved with pre_mcp_call before this fix) no longer disappears from the dropdown; the option renders with a 'not supported by <provider>' note so the user knows to pick another. * fix(guardrails): correct audited hook lists, prune stale modes on provider switch, clean form lint Audited every get_supported_event_hooks classmethod against the hooks each guardrail's own tests exercise and its handler methods. Five were too narrow and their tests caught it in CI: rubrik gains pre_call, presidio gains during_call and pre_mcp_call, prompt_security, onyx and qualifire gain during_call. The remaining classes match either their original __init__ declarations or their exercised modes exactly. Cursor review fixes: the Add form now drops selected modes the new provider does not support when the user switches providers, so a pre_mcp_call selection cannot ride along into a provider that rejects it at save; the edit form handles list-shaped stored modes instead of treating mode as always a string. Extracted shared toModeArray and getSupportedModesForProvider helpers into guardrail_info_helpers so both forms use one implementation, typed the remaining any usages in both forms, removed nested ternaries, and committed the ratcheted-down eslint metrics and pruned suppressions * fix(proxy): reserve tiered output at the higher reasoning rate Some tiered Dashscope models price reasoning output above standard output (output_cost_per_reasoning_token > output_cost_per_token). The reservation charged all output at the standard rate, so a reasoning-heavy request reserved too little and concurrent calls could exceed the budget before reconciliation. The reasoning share is unknown before the request runs, so reserve every output token at the higher of the two configured rates, for both tiered and flat pricing. Co-authored-by: Cursor <cursoragent@cursor.com> * refactor(ui): convert user agent and per-user usage charts to shadcn/recharts (#32725) * refactor(ui): convert user agent and per-user usage charts to shadcn/recharts Swap the tremor BarChart import for the shared shadcn/recharts wrapper in user_agent_activity.tsx (DAU/WAU/MAU charts) and per_user_usage.tsx (usage distribution histogram). All chart props are unchanged; the wrapper exposes the same tremor prop surface with matching defaults. Extend user_agent_activity.test.tsx and add per_user_usage.test.tsx with parity assertions on the real recharts SVG output: bar series per category, stacked x positions, resolved fill colors, axis bucket labels, legend text, and value formatter output on axis ticks. Remove the dead ResizeObserver polyfill in user_agent_activity.test.tsx now that the scoped global mock in tests/setupTests.ts renders charts, which also lowers the no-explicit-any metric by one. * test(ui): harden bar x-position parsing against recharts path format * feat(ui): working Test Connection for the complexity auto router The consolidated auto-router tab dropped the Test Connection button because the shared prepareModelAddRequest helper returns an empty array for an auto router (it has no model_mappings), so the caller crashed destructuring result[0].litellmParamsObj. That is the crash in #31590 and the open PR #31794. #31794 only silenced the crash by pointing the test at auto_router/complexity_router, which is not a provider model, so the /health/test_connection health check (a real litellm.ahealth_check completion) would still error. Bring the button back and make it meaningful: an auto router dispatches to saved model groups, so Test Connection now probes those directly. It builds a deduped target list from the configured tiers (tiers sharing a model group collapse to one probe) plus the embedding model when semantic keyword matching is on, then runs a live /health/test_connection against each and shows per-target pass/fail. This never touches prepareModelAddRequest, so the original destructure crash cannot recur. Scope is the recommended complexity router only; the to-be-deprecated semantic router is untouched. No backend changes. Supersedes #31794. Resolves #31590. * refactor(ui): extract a shared CopyButton and fix the sidebar copy confirmation (#32945) * fix(ui): show sidebar copy confirmation only on a successful write The sidebar account menu's copy button switched to the checkmark synchronously, before the clipboard write settled, so it confirmed a copy that never happened when navigator.clipboard was undefined on non-secure origins or when writeText rejected. The handler now guards navigator.clipboard, awaits the write, and flips to the checkmark only on success Also updates the header accent emoji in the same menu * refactor(ui): extract a shared CopyButton for the sidebar account menu The copy-icon-to-checkmark pattern was hand-rolled in several places, including the sidebar account menu whose private copy button held the false-confirmation bug. Extract a single canonical CopyButton into components/shared, built on the Button primitive with a guarded and awaited clipboard write so the checkmark appears only on a real success, and have SidebarAccountMenu consume it The success and failure-mode coverage now lives in the shared component's own test; the sidebar test keeps one case asserting the email row is wired to it * fix(ui): probe auto-router tiers via real proxy routing, not /health/test_connection Live testing showed the first cut was broken: /health/test_connection merges {...configParams, ...requestParams}, so passing the public model_group name as the request model overrode the resolved provider model and every tier failed with "LLM Provider NOT provided". The frontend only has the public group name, not the underlying litellm_params, so it cannot build the request that endpoint needs. Switch to testing each model group the way production actually routes it: send a minimal request to /v1/chat/completions (or /v1/embeddings for the embedding model) by public group name through the shared apiClient. The router resolves the group, credentials, and provider itself, so a green row means the tier is genuinely reachable. Verified live: voyage embedding returns 200, a tier with a bad key returns the real provider auth error. Also address Greptile feedback: rows now update progressively as each probe settles instead of all at once, and TIER_ORDER is derived through a `satisfies Record<keyof ComplexityTiers, null>` guard so adding a tier without listing it is a compile error. * fix(completion): forward aws credential kwargs into litellm_params so the responses bridge keeps WIF auth Chat-completions requests to responses-only Bedrock Mantle models are bridged to the Responses API, but completion() forwarded only aws_bedrock_project_id into get_litellm_params, so aws_role_name, aws_web_identity_token, aws_session_name and the other SigV4 credential kwargs never reached sign_request and botocore fell back to the default credential chain ("Bedrock Mantle auth failed: no Bearer token and no usable AWS credentials"). Forward the whole AWS credential kwarg family, extracted from the OPTIONAL_KWARGS_KEYS set get_litellm_params already supports. * refactor(ui): convert entity usage and usage page charts to shadcn/recharts (#32729) * refactor(ui): convert entity usage and usage page charts to shadcn/recharts Swap the tremor BarChart/DonutChart render sites in EntityUsage, SpendByProvider, TopKeyView, TopModelView, KeyModelUsageView and UsagePageView to the shared shadcn/recharts wrappers. Convert the two sole-chart Daily Spend cards and the KeyModelUsageView card to the shadcn Card primitives. Close the donut parity gap with strictly additive optional DonutChart props: showLabel/label render a center total (tremor showed valueFormatter(sum) by default) and startAngle/endAngle forward to the Pie so both provider donuts keep tremor's clockwise-from-12 layout. Defaults preserve the previous wrapper behavior. DailyData and two site-local row types move from interface to type alias so they satisfy the wrappers' Record<string, unknown> constraint; interfaces lack implicit index signatures. Tests now assert on real recharts output: bar/sector counts, cyan fills, axis labels, donut center totals, and the TopKeyView bar-click drill-down into the key info modal. The dead tremor chart mocks in UsagePageView.test.tsx are removed and lint metrics/suppressions are regenerated for the dropped tremor imports. * fix(ui): compute donut center label only when shown and assert Model Usage renders as a card title * refactor(e2e): bucket rate limits, budgets, and spend tracking under quota_management * fix(e2e): name the route the spend_calculate registry cell actually exercises * refactor(e2e): move budgets and spend_tracking suites under quota_management * test(e2e): cover key rpm/tpm rate limiting, window reset, and pacing headers * test(e2e): assert the tpm block at its exact token crossing instead of a call-count heuristic * test(e2e): fail the rpm reset test when the limiter resets early * test(e2e): name the tpm window deadline's latency margin * refactor(e2e): model the tpm spend loop's two outcomes as values * test(e2e): source the ratelimit suite's model from E2E_CHEAP_ANTHROPIC_MODEL * feat(guardrails): add pre_mcp_call support to Content Filter (#32936) * feat(guardrails): add pre_mcp_call support to Content Filter * test(guardrails): cover canonical MCP key gate under pre_mcp_call mode * fix(guardrails): scan MCP arguments per value and gate mixed-mode scans by call type * fix(guardrails): cap MCP argument scan depth and register the walker with the recursion detector * test(guardrails): update LIT-4226 UI settings tests for content filter pre_mcp_call support * fix(guardrails): use builtin generics in MCP scan annotations to satisfy strict-rule budget Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> * fix(bedrock): allow bedrock-mantle:CreateInference in the web identity session policy * fix(ui): drop max_tokens from the auto-router connection probe max_tokens=1 makes reasoning models (o1/o3/...) return a 400 "max_tokens reached" because reasoning tokens count against the cap, so a reachable reasoning tier showed a false failure in Test Connection. Live-verified: o3 400s with the cap and succeeds without it. Extract the request shape into a pure buildModelGroupTestRequest and cover it with a test asserting the chat body carries no max_tokens (or max_completion_tokens), so this regression is caught in unit tests instead of only against a live reasoning model. * test(main): assert the responses bridge forwards static aws keys as well as web identity params * bump: litellm-proxy-extras 0.4.75 -> 0.4.76 (#32957) * feat(proxy): add expires filter to GET /key/list (#32953) * feat(proxy): add expires filter to GET /key/list Add an opt-in expires query param to GET /key/list so callers can fetch only expired or only active keys without paginating every page and filtering client-side. 'expired' matches keys whose expires is in the past (NULL expires excluded); 'active' matches keys that never expire or expire in the future. Omitting the param preserves existing behavior for every caller. An unrecognized value returns HTTP 400 rather than silently returning all keys. The filter is pushed to the database via the existing Prisma where builder so callers avoid pulling the full key table into application memory. Resolves LIT-3387 * refactor(proxy): declare VALID_EXPIRES_FILTER_VALUES before its first use * refactor(ui): colocate the usage view, keeping the shared usage components (#32952) Split for the usage (UsagePage) segment. Most of the folder is the usage page's own view, but four pieces are reused elsewhere and stay in @/components/UsagePage: TopKeyView (old-usage), KeyModelUsageView and value_formatters (activity_metrics), and the shared types (activity_metrics, chartUtils). The other 21 files move into usage/_components, preserving the folder structure. The external consumers import only the retained files, so they are untouched. The moved files' imports of the retained files become @/components/UsagePage paths, other escaping relative imports are absolutized, and lint suppressions are re-keyed for moved files only. No behavior change. * chore: update Next.js build artifacts (2026-07-11 23:35 UTC, node v20.20.2) (#32960) * test(e2e): cover model-aware mid-conversation system handling on Bedrock Invoke /v1/messages * docs(github): add QA runbook section to the PR template * docs(github): scope the QA runbook to tests/e2e edits and add example checklists * docs(github): shape QA runbook examples as node id plus behavior bullets * refactor(ui): convert projects page chart to shadcn/recharts (#32722) * fix(xecguard): use StandardLoggingGuardrailInformation in logging hook (#32911) XecGuard's async_logging_hook wrote a bare dict to standard_logging_object["guardrail_information"] while the typed contract is Optional[List[StandardLoggingGuardrailInformation]]. Readers that iterated the field walked dict keys, raised on info.get, or silently dropped the entry from guardrail usage tracking and spend-log writes Construct the typed entry and append it to the existing list or create a new one, matching the shared helper pattern. Record the configured guardrail name instead of a hardcoded "xecguard" and pass the GuardrailEventHooks enum for guardrail_mode * feat(ui): adopt openapi-react-query ($api) and convert useCustomers (#32949) * feat(ui): adopt openapi-react-query and convert useCustomers to $api Add openapi-react-query and expose $api = createQueryClient(fetchClient) alongside fetchClient. Rewrite useCustomers as $api.useQuery("get", "/customer/list", {}, { enabled, select }), which derives the query key from method + path (dropping the hand-written createQueryKeys entry and the manual key) and forwards the request signal for cancellation. The response type still flows from schema.d.ts as CustomerResponse[]. Tests assert the path, the admin/token enabled gate, and the empty-body select fallback. * test(ui): read the last render's options in useCustomers helper The lastCallOptions helper was named for the last call but read mock.calls[0]. Harmless while each test renders once, but it would silently assert against first-render options if a test ever re-renders. Read the final call instead. * refactor(ui): colocate the mcp-servers view, keeping the shared mcp_tools surface (#32968) * refactor(ui): colocate the usage view, keeping the shared usage components Split for the usage (UsagePage) segment. Most of the folder is the usage page's own view, but four pieces are reused elsewhere and stay in @/components/UsagePage: TopKeyView (old-usage), KeyModelUsageView and value_formatters (activity_metrics), and the shared types (activity_metrics, chartUtils). The other 21 files move into usage/_components, preserving the folder structure. The external consumers import only the retained files, so they are untouched. The moved files' imports of the retained files become @/components/UsagePage paths, other escaping relative imports are absolutized, and lint suppressions are re-keyed for moved files only. No behavior change. * refactor(ui): colocate the mcp-servers view, keeping the shared mcp_tools surface * docs(github): add Final Attestation and per-test sanity-check step to QA runbook * refactor(ui): convert endpoint usage charts to shadcn/recharts (#32723) * refactor(ui): convert endpoint usage charts to shadcn/recharts Adds a LineChart wrapper to the shared charts kit, mirroring the BarChart/AreaChart composition with connectNulls and curveType props, and converts EndpointUsageBarChart and EndpointUsageLineChart from tremor to the shared wrappers. Both endpoint chart tests now assert on real recharts SVG output instead of tremor mocks. * refactor(ui): drop unused endpointData prop from EndpointUsageLineChart * fix(ui): point endpoint chart test type imports at the UsagePage types alias after colocation move * fix(auto_router): filter embedding models out of tier selects, require all tiers, add inline validation The Add Auto Router complexity tab let chat models fill the embedding-model slot (and vice versa) since neither dropdown filtered on ModelGroup.mode, and submit only required at least one of the four tiers instead of all four. Adds getMissingTiersError alongside the existing getSemanticConfigError, and highlights unfilled tier/embedding selects inline once a submit attempt fails. * fix(model-cost-map): anchor the bedrock-claude-ids routing rule to the start of the id * fix(proxy-auth): deny provider-wildcard access inferred through an unrecognized model namespace * fix(auto_router): reset inline validation errors when switching router type * fix(auto_router): flag name field and tier fields together on empty submit Clicking Add Auto Router with the name empty returned early with only a toast, so blank tier selects never got their inline error state. The empty-name branch now sets showValidationErrors and triggers antd validation on the name field, so every unfilled mandatory field is flagged at once. Adds a regression test for the tab component. * feat(mcp): mint gateway-bound envelope at the token endpoint for dcr_bridge oauth_delegate * feat(mcp): seal the authorizing key hash in the dcr_bridge envelope The mint bound only user_id/server_id into the envelope, which gave admission no way to reload the caller's key and enforce its current restrictions. Seal the hashed authorizing key instead (a one-way digest, not a usable credential), so admission reloads the live UserAPIKeyAuth by it and the key's team/org/tool permissions and revocation apply per request. Extract the token endpoint's key resolution into a shared _resolve_active_litellm_key so the per-user token store (user_id) and the bridge mint (key hash) derive from one active-key-gated path, and fail the mint closed with invalid_request when no active key accompanies the request. * fix(mcp): return 502 not KeyError when a bridge upstream response lacks access_token The eager access_token = token_response["access_token"] extraction ran before the dcr_bridge branch, so a missing upstream access_token raised an unhandled KeyError and _bridge_grant_from_token_response's nil guard (which maps to a clean 502) was dead code. Move the extraction onto the non-bridge result path so the bridge branch reaches its 502 guard. * fix(mcp): let a keyless-user active key mint a bridge envelope _resolve_active_litellm_key gated on _active_key_user_id, which returns None both for blocked/expired keys AND for valid keys with no user_id, so a team-scoped or service-account key was wrongly rejected with invalid_request at bridge token exchange. Split the active-state gate (_key_is_active: blocked/expiry only) from the user_id extraction; the mint seals the key hash, not the user, and admission already handles a keyless-user key. The per-user token store still gets no user for such a key, as there is none to key a stored credential by. * style(mcp): use X | None annotations on the touched key-resolution helpers The keyless-user fix moved these signatures, so their pre-existing Optional[...] annotations counted against the diff and tripped the UP045 strict-budget gate. Modernize the four touched return annotations to the X | None form the gate wants; runtime behavior is unchanged. * fix(mcp): coerce numeric expires_in and make the active-key check total Two correctness gaps in the bridge mint. _bridge_grant_from_token_response only accepted an int expires_in, dropping a float (3600.0) or numeric-string ('3600') lifetime to None so the envelope fell back to its 1h cap and could outlive a shorter-lived upstream token; coerce it to a positive int (bool excluded). And _key_is_active called datetime.fromisoformat on the str|datetime expires outside the resolver's try, so a malformed stored expiry raised an unhandled 500 instead of the fail-closed invalid_request; it now fails closed (inactive) on an unparseable expiry. Regression tests cover int/float/string/bool coercion, the short-float TTL, and the malformed-expiry fail-closed path. * fix(mcp): harden the bridge token mint (multi-lens review pass) Findings from a full adversarial review of the mint path across security, correctness, error-handling, concurrency, and OAuth-protocol dimensions. - expires_in coercion is now total: int(float(...)) can raise OverflowError on Infinity / a giant numeric string, which escaped the ValueError/TypeError catch and 500'd the token endpoint. Unified to catch OverflowError too. - Resolve the litellm identity BEFORE exchanging the single-use upstream code, so a missing or transiently-unresolvable identity fails closed with invalid_request without burning the code (the mint re-resolves via a cache hit). - The no-identity failure is now an RFC 6749 5.2-shaped invalid_request (JSONResponse, top-level error, no-store) instead of a detail-wrapped HTTPException, matching the BYOK OAuth endpoint. - EnvelopeTooLarge (upstream token too big to seal) surfaces a 502, not a 500. - The upstream refresh_token is no longer sealed into the envelope: the edge never consumes it, so it was dead weight embedding a long-lived upstream credential in the client bearer and enlarging the envelope; refresh is a follow-up (a dedicated refresh-envelope). Security review found no exploitable defect (forgery, cross-server/user replay, leakage, confused-deputy all closed). Regression tests cover the OverflowError, the code-not-burned path, the RFC-shaped error, the 502, and the dropped refresh. * fix(mcp): close the burn-before-check gate for both grants and validate master_key first Follow-up to the pre-exchange identity gate, which I had only added to the authorization_code branch and which left the master_key check inside the mint (after the upstream exchange) - so the very burn-then-fail pattern it was meant to prevent still applied to refresh_token grants and to a misconfigured gateway. - Hoist a single pre-exchange gate above the upstream call that covers BOTH grant types: it fails closed (invalid_request) on an unresolvable litellm identity and 500s on an unset master_key BEFORE the single-use code or refresh token is exchanged/rotated, so a bad key or a misconfigured gateway never burns the upstream credential. - Report expires_in from the envelope JWT's own second-truncated exp (rounding the elapsed portion up) instead of the raw expires_at - now delta, so the client is never told the bearer is valid past the ~1s point admission already expires it. Regression tests assert the upstream exchange is never called on the no-identity refresh grant and the master_key-unset path, and that the reported expires_in does not overstate the JWT exp. * refactor(mcp): make the bridge delegate mint a phased failures-as-values pipeline The dcr_bridge oauth_delegate token mint validated its preconditions in two places: a pre-exchange guard inside exchange_token_with_server (master_key set, resolvable litellm identity) and an authoritative re-check inside the post-exchange _mint_bridge_delegate_token_response. Keeping the two in step by hand is what kept producing the same class of finding: a precondition guarded on one grant branch but not the other, master_key checked after the exchange on one path, identity resolved twice, and each failure raising an ad-hoc HTTPException with its own status and body shape. Model the mint as three phases whose failures are values. _prepare_bridge_mint runs before the exchange, checks every precondition once (master_key, then identity), and returns either a frozen _BridgeMintReady carrying the resolved key hash and the master-key-derived envelope keys, or a _BridgeMintError literal. Because every precondition lives in prepare, and prepare runs before the upstream POST, no failure can burn the single-use code or rotate a refresh token, for either grant type, by construction rather than by a guard we have to remember to keep in sync. _finish_bridge_mint runs after the exchange and has no preconditions left that can fail; its only failure values are properties of the upstream response itself (no usable access_token, or a token too large to seal). One mapper, _bridge_mint_error_response, turns each _BridgeMintError into an RFC 6749 section 5.2-shaped body with a status truthful about where the failure is (400 for the caller, 500 for gateway config, 502 for the upstream), with an exhaustive match plus assert_never so a new failure mode cannot be added without a matching status. Behavior is unchanged for the client. Every failure that previously raised now returns the same status as an OAuth error body, which is the correct token-endpoint contract; the three tests that asserted a raised HTTPException now assert the returned response. _exchange_for_bridge_server additionally asserts the identity resolver is awaited exactly once for a bridge server and never for a non-bridge one. * fix(mcp): let the bridge envelope report expires_in 0 at the jwt exp boundary _finish_bridge_mint floored the reported expires_in at 1. Admission expires the envelope against the JWT's second-truncated exp, so when the mint lands in the same second that exp falls on (a sub-second upstream lifetime, for instance), the true remaining life is 0 and reporting 1 tells the client the bearer lives one second past the point admission already rejects it. Floor at 0 instead so the reported lifetime never overstates the exp; the value still cannot go negative. The regression pins the boundary directly: minting at now=100.25 with a 1s upstream token seals exp=101, and the reported expires_in is max(0, 101 - ceil(100.25)) = 0. Under the old floor of 1 it reads 1, so the test fails on that mutation. Also drops the unused mcp_server parameter from _prepare_bridge_mint; identity and key derivation there never referenced the server. * refactor(mcp): make bridge-mint resolvers return tagged unions so status is truthful by construction Three findings landed together, all one defect: a resolution step crushed several distinct outcomes into a single None or a silent default, so the mint's error mapper could not tell them apart and assigned the wrong status. Identity resolution mapped a database outage to the same None as a missing credential, which the mint reported as 400 invalid_request, blaming the caller for a gateway outage while admission statuses the same outage 503/500. Lifetime coercion mapped an explicit non-positive expires_in to the same None as an absent one, so an upstream token the IdP reports as already dead was sealed into an hour-long envelope. And the refresh_token grant was run through the upstream exchange (which can rotate the client's upstream refresh credential) and its result then discarded, even though a bridge server seals no refresh_token and the client never holds one to present. Rather than add a mapping branch per finding, the fix changes the return types so a wrong status is not representable. Each resolution step now returns a precise tagged value instead of None: identity resolution returns a _ResolvedKey or one of no_active_key / unavailable / unresolvable, classified the same way admission's _reload_admitted_key classifies the same conditions; upstream-lifetime classification returns a positive number of seconds, "unspecified" (absent or unparseable, which the envelope caps), or "expired" (a parseable non-positive value, an already-dead token); and upstream-grant validation returns a typed grant or one of no_access_token / expired_lifetime. Thin exhaustive mappers (match plus assert_never) lift each vocabulary into one bridge-mint taxonomy of eight named failures, and a single _bridge_mint_error_response gives each its truthful RFC 6749 §5.2 status: 400 for the caller's missing credential or an unsupported grant, 503 for a transient auth-DB outage, 500 for a gateway that cannot resolve identity or is not configured, and 502 for an upstream response with no usable token, an already-expired lifetime, or a token too large to seal. Adding a failure mode now requires a new literal and a match arm the type checker forces, so the class of wrong-status bug cannot recur silently. The refresh_token grant is rejected in _prepare_bridge_mint before the exchange with unsupported_grant_type, so it can never rotate or consume the client's upstream refresh credential; renewal is re-running authorization_code, as the sealed refresh_token=None already intends. An absent or unparseable expires_in still mints a capped envelope (the by-design behaviour for an upstream that omits the field); only an explicitly-dead lifetime is rejected. Tests cover the resolver's three failure classes (including a real connection-error outage and a missing prisma_client), the mint statuses for each (503 before the upstream exchange, 500, 502 on an expired upstream lifetime, and a capped mint on an unknown one), and the refresh-grant rejection before any exchange. The three findings are mutation-checked: reverting each fix turns its regression test red. * fix(mcp): treat a positive sub-second upstream lifetime as alive, not expired _classify_upstream_lifetime decided "expired" from int(float(expires_in)), which truncates toward zero, so a positive fractional lifetime in (0, 1) became 0 and was misread as already elapsed. That rejected the mint with 502 in _finish_bridge_mint after the single-use upstream code had already been consumed, even though the upstream reported a positive remaining lifetime. Decide expired on the parsed numeric value rather than its truncated int, so only a genuinely non-positive value is expired. The envelope works in whole seconds and cannot represent a sub-second lifetime, so a positive value that truncates to 0 clamps up to the 1s floor instead of being rejected. Values >= 1 still truncate toward zero so the envelope never claims more life than the upstream stated, and NaN / Infinity / oversized input still read as unparseable ("unspecified"). Regression covers the classifier (0.5 and 0.001 clamp to 1, 1.9 truncates to 1, -0.5 stays expired) and the mint (a 0.5s upstream lifetime mints a 200 envelope rather than a 502); reverting to the truncate-then-check reddens both. * fix(auto_router): inline error for missing LLM classifier model Selecting the LLM classifier without picking a model only surfaced a toast on submit; the classifier model select now gets the same red outline and helper text as the tier and embedding selects once a submit attempt has failed. * build(dev-env): add make bootstrap and unprovisioned-checkout preflight to pre-commit * feat(router): random-pick multi-model complexity tiers (#32967) * feat(router): random-pick multi-model complexity tiers Tier pools already make sense without adaptive; stop pinning lists to index 0 and shuffle within the classified tier instead. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(ci): format complexity router config Co-authored-by: Cursor <cursoragent@cursor.com> * fix(ci): use PEP 585 types for tier pools Co-authored-by: Cursor <cursoragent@cursor.com> --------- Co-authored-by: Cursor <cursoragent@cursor.com> * fix(xecguard): sanitize scan result before recording it for logging (#32935) * chore: keep it brief * chore: keep it brief * docs(readme): point developer-mode setup at make bootstrap * chore: keep it concise * feat(router): add Router(plugins=[...]) routing-plugin pipeline (#32972) * feat(router): add Router(plugins=[...]) routing-plugin pipeline Runs a sequence of user-supplied plugins before the routing decision is made. Each plugin reads/mutates a RoutingContext (messages, candidate models, metadata, signals); the narrowed candidate list is enforced when picking a deployment, raising rather than silently falling back if a plugin narrows to zero candidates. Prototype for the routing-plugin pipeline discussed in #32168. * fix(router): use ruff-modern typing, add raw/structured messages to RoutingContext - Use dict/list/X|None instead of Dict/List/Optional in new code, staying within the ruff strict-rule budget ratchet - Extract the guardrail-translation message normalization ComplexityRouter already had into a shared resolve_structured_messages() helper (litellm_core_utils/prompt_templates/factory.py), reused by ComplexityRouter and the new routing-plugin pipeline instead of duplicating it - RoutingContext now exposes both raw_messages (as received) and structured_messages (normalized across chat completions / Anthropic messages / Responses API), mirroring CustomGuardrail.apply_guardrail's pattern, per review feedback on #32972 - Add direct unit tests for _run_routing_plugins and _filter_by_routing_plugin_candidates (router_code_coverage gate requires every router.py function be called by name somewhere in tests/) * fix(test): rename to test_router_routing_plugins.py router_code_coverage.py's AST scanner only inspects test files whose filename contains the substring "router" -- test_routing_plugins.py doesn't match (routing != router), so it silently skipped this file and flagged _run_routing_plugins/_filter_by_routing_plugin_candidates as untested despite the direct unit tests added for them. * fix(router): fail closed when plugins are configured but the resolved routing path can't run them Router.completion() (and other sync entry points) resolves deployments via the synchronous get_available_deployment(), which never runs async_pre_routing_hook and therefore never runs the routing-plugin pipeline. async_get_available_deployment() itself falls back to that same synchronous method for routing strategies without an async-native selector (e.g. legacy "usage-based-routing" v1). Both paths would let a policy plugin (e.g. a deny-all rule) be silently bypassed. Raise instead of silently proceeding when self.routing_plugins is configured and the sync path is reached, since applying the pipeline to every selector path is a larger change out of scope for this PR. Per review: https://github.com/BerriAI/litellm/pull/32972/changes/BASE..bdfb583c2c6f8df10004fb249e11629d41ce71fa#r3565373303 * feat(router): soft-floor adaptive mode for complexity router (#32947) * feat(router): soft-floor adaptive mode for complexity router Let complexity_router_config.adaptive=true Thompson-sample across the union of tier pools with a tier-distance penalty, and wire the existing adaptive post-call bandit so mis-tiered requests can still recover. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(router): reattach adaptive hooks for hybrid complexity Finalize was wiping every AdaptiveRouterPostCallHook and only re-registering standalone auto_router/adaptive_router deployments, so complexity adaptive=true never received bandit updates. Co-authored-by: Cursor <cursoragent@cursor.com> * chore(router): drop unnecessary hybrid docstrings Co-authored-by: Cursor <cursoragent@cursor.com> * fix(router): attribute adaptive feedback Credit user reactions to the model that produced the previous response while keeping current-response signals on the serving model Co-authored-by: Cursor <cursoragent@cursor.com> * fix(router): tune hybrid cold defaults Use the cost-weighted policy that beat equal-pool complexity in the full bakeoff, and make the committed harness compare identical tier pools Co-authored-by: Cursor <cursoragent@cursor.com> * fix(router): preserve hybrid cold quality floor Sample only unobserved models in the classified tier until feedback exists, then apply adaptive scoring without mis-penalizing models shared across tiers Co-authored-by: Cursor <cursoragent@cursor.com> * fix(router): bound feedback context cache Cap retained session feedback so unique session IDs cannot exhaust router memory Co-authored-by: Cursor <cursoragent@cursor.com> * fix(router): preserve exhaustion signals Include tool-result exhaustion in adaptive feedback and clear strict lint regressions blocking CI Co-authored-by: Cursor <cursoragent@cursor.com> * refactor(router): remove stale owner cache Remove obsolete attribution state, tighten the embedded router type, and keep the test diff focused on adaptive behavior Co-authored-by: Cursor <cursoragent@cursor.com> * refactor(router): centralize hook cleanup Use the callback manager to discover and remove adaptive hooks across every registered callback list Co-authored-by: Cursor <cursoragent@cursor.com> --------- Co-authored-by: Cursor <cursoragent@cursor.com> * fix(responses): continue MCP gateway tool turns from the final response and surface failures When a /responses request uses a hosted MCP tool (server_url: litellm_proxy/<label>) with store=true and the model calls a tool, the gateway auto-executes the tool and streams one logical response stitched from several upstream responses: an interim response whose only output is the function_call, then the post-tool answer B1 (correctness): every streamed event was pinned to the first round's response id, i.e. the interim response that carries the function_call but no tool output. The client then continued the next turn from that dangling response and the provider rejected it with "No tool output found for function call <id>", which on the streaming path surfaced as a silent empty completion. The fix adopts each auto-execute round's own response id (the cached id is reset when a follow-up round starts) so the client continues from the final round, whose stored input chain includes the function_call_output B2 (robustness): initial and follow-up call failures were swallowed; the stream emitted the mcp_list_tools discovery events and then closed with HTTP 200 and no output and no error. The fix stashes the failure, makes the initial call eagerly in aresponses_api_with_mcp so a pre-stream failure re-raises as a real 4xx before any SSE bytes are written, and emits a terminal error event when a follow-up call fails mid-stream Adds regression tests covering continuation exposing the final round's response id rather than the interim tool-call id, a follow-up failure emitting a terminal error event, and an initial-call failure being stashed for eager re-raise * ci(ui): report only error-level knip findings in CI (#32971) * feat(batches): track cost for unmanaged Bedrock batches, generalize the flag (#32315) * feat(batches): track cost for unmanaged Bedrock batches, generalize the flag CheckBatchCost skipped Bedrock batches whose unified_object_id is a raw model-invocation-job ARN, the same root cause previously fixed for unmanaged Vertex batches. Bedrock batches embed the model name in their s3:// input file name instead (litellm-bedrock-files-{model}-{uuid}.jsonl), so the same routing mechanism now derives the model from that layout and matches it to a configured bedrock deployment. track_unmanaged_vertex_batch_cost is renamed to track_unmanaged_batch_cost since two providers now share this mechanism. * fix(batches): parse Bedrock batch output and price with deployment model name Bedrock model-invocation-job results use modelOutput/error rows and short internal model ids that are not in the cost map, so unmanaged batch cost tracking logged tokens but $0 spend. Use deployment model name for pricing and add regression tests. Co-authored-by: Cursor <cursoragent@cursor.com> --------- Co-authored-by: Cursor <cursoragent@cursor.com> * fix(guardrails): walk custom_tool_call_output items in _content_utils (#32969) * fix(guardrails): walk custom_tool_call_output items in _content_utils * Change _OUTPUT_ITEM_TYPES to Frozenset type * fix(guardrails): use builtin frozenset generic for _OUTPUT_ITEM_TYPES annotation Frozenset is not a defined name (typing exports FrozenSet, the builtin is frozenset), so module import raised NameError and broke every proxy test suite. The builtin generic is valid on the supported python floor (3.10) and keeps the UP006 ruff-strict budget at its ceiling, which the typing alias would exceed * fix: show and allow editing team model aliases after team creation (#33047) * refactor(ui): rename OldTeams component file to Teams * fix: show and allow editing team model aliases after team creation * fix(ui): mark team model_aliases as nullable to match the prisma schema * fix(ci): bump pillow to 12.3.0 to resolve osv-scan CVEs (#33093) * fix(proxy): track unauthenticated pass-through requests in spend logs (#32410) Pass-through endpoints configured with auth=false reach the cost-tracking callback with no key/user/team/end-user, so _should_track_cost_callback returned False and the spend-log write was skipped, leaving the request out of request/usage logs. Track pass-through call types even when unauthenticated so the SpendLog row is still written. Co-authored-by: Mubashir Osmani <mubashir@berri.ai> Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * feat(lasso): send source.type=litellm for Used By attribution (#33090) Co-authored-by: Or Gershoni <org@lasso.security> * feat(mcp): generalize the bridge envelope identity to a key_hash or user_id subject The scripted two-header client mints under a virtual key it presents at the token endpoint (key_hash), but the interactive DCR client authenticates via SSO at the bridged authorize, which yields a user, not a key. Make EnvelopeIdentity a discriminated subject (subject_type key_hash | user_id) with key_hash_identity / user_identity constructors, and dispatch admission on it: a key_hash reloads the key, a user_id reloads the user and admits them as themselves (user-level budget and SCIM enforced via the same centralized gate; no team bound, since a user belongs to many teams or none). The interactive producer that mints a user_id envelope lands in the follow-up commit. * feat(mcp): interactive SSO sign-in for dcr_bridge oauth_delegate DCR clients Completes the oauth_delegate bridge for real DCR clients (Claude Code, Claude Desktop), which send no litellm key and cannot use the scripted two-header path. On the short-circuit bridge arm the gateway now captures the SSO-authenticated litellm user from the browser session at /authorize and seals it into the OAuth state; at /callback it seals that user plus the upstream code into a gateway authorization code the client echoes back; at /token it recovers the user, exchanges the real upstream code, and mints a user-subject envelope. The user identity captured in the browser thus rides to the back-channel token call with nothing stored server-side, and admission opens the envelope under that user. The scripted key_hash path is unchanged (raw upstream code, key from the request); without a session the browser is sent through login first. * fix(mcp): classify the user-subject reload's errors like the key path (503 outage, 401 missing) _reload_admitted_user mirrored only part of _reload_admitted_key's error contract: it caught ProxyException and HTTPException but had no arm for anything else, so a transient DB outage surfaced as an opaque 500 instead of the retryable 503 the key path guarantees, and a missing user surfaced as a 500 too. The missing-user case is the subtle one: get_user_object raises a bare Exception for a deleted user (not a ProxyException like get_key_object does for a missing key), so the ProxyException/HTTPException clause never caught it and the user_object-is-None branch it was supposed to hit is unreachable on the production path. Add the same except-Exception arm the key path uses, with the one deliberate difference the differing get_user_object contract requires: a database-service-unavailable error still raises the retryable 503, while a missing user or any other non-outage resolution failure fails closed as a 401 rather than propagating as a 500. The regression tests now drive the real behavior (get_user_object raising) rather than a None return that never happens in production, and cover both the 503 outage and the 401 missing-user paths. * fix(mcp): admit a user-subject envelope with the user's own MCP object permission _reload_admitted_user returned a bare UserAPIKeyAuth(user_id=...), so the shared get_allowed_mcp_servers found no key/team/object-permission grants and an interactive SSO client could admit successfully yet see zero tools on a normal (allow_all_keys=False) server. The key path returns the full key record whose object permission drives that computation; the user path dropped it. Resolve the user's own MCP object permission and put it on the returned auth, so the same get_allowed_mcp_servers the key path uses grants the user their litellm-granted servers and access groups. This reuses get_object_permission (the id-to-grants resolver keys and teams already use) and does not duplicate any permission logic; get_user_object does not load object_permission, so it is resolved from the user's object_permission_id the same way the key and team paths do. Only the user's own object permission is bound. A UserAPIKeyAuth carries a single team_id while a user may belong to many teams, so team-inherited MCP grants for a user are a follow-up: they need a many-teams union get_allowed_mcp_servers does not do off one auth object, and faking one here would be the kind of half-measure that spawns more bugs. Tests cover the user's object permission riding onto the admitted auth, and the existing admit/SCIM/missing-user/503 cases still hold. * fix(ui): respect litellm_key_header_name in BYOK credential save and workflow runs fetches (#33103) * refactor(ui): standardize debounce waits behind shared DEBOUNCE_WAIT_MS constant (#33040) * feat(ui): rebuild the Virtual Keys table on the shared DataTable (#32991) * feat(ui): rebuild the Virtual Keys table on the shared DataTable Replaces the hand-rolled Tremor table and bespoke toolbar/pagination on the admin Virtual Keys page with the shared DataTable: server-side sort, paginate, and filter, a sticky scrolling body, a search plus column-visibility plus filters toolbar, a right-side filter drawer, and a rows-per-page footer. A page header with the existing key icon carries the Create New Key action. Adds reusable, shadcn-default building blocks for the tables migrating onto the DataTable next: shared IdentityCell, ModelsCell, and SpendBudgetCell in shared/table_cells, plus a shared PageHeader. The models cell reveals overflow in a hover tooltip and the spend/budget cell uses the Meter primitive. All data and domain logic is preserved, including the useKeys query, team and org alias resolution, the user popover, and the KeyInfoView detail swap. The rich async Team/Org/Alias filters move into the drawer, and the toolbar search maps to the key-alias substring search. Status now also reflects key expiry alongside blocked and SCIM-blocked. The VirtualKeysTable tests are updated to the new markup and extended with focused coverage for each new shared cell * fix(ui): address Virtual Keys redesign review feedback Fold the status badge into the clickable Key cell and drop the separate Status column so a key's alias, secret, and status read as one unit. The Key cell is now the single click target that opens the key detail; the whole-row click is removed Migrate the filter drawer off AntD to shadcn. A new Combobox composed from Popover and Input backs the Team, Organization, and Key Alias filters, keeping search and the alias infinite-scroll Show $0.00 for zero spend instead of a hyphen, and extend the shared DataTable with badge, chips, and meter skeleton shapes so the loading state matches the loaded cells (status pill, model chips, spend meter) rather than uniform bars Fix key sorting: the Key column sent its column id "key" as sort_by, which /key/list rejects with 400. It now sorts by the backend field key_alias * fix(ui): use the shadcn base combobox and refine the keys filters and skeletons Replace the hand-rolled filter combobox with the supported shadcn Base UI combobox (ui/combobox, added via the CLI and reused through a small SearchSelect wrapper). Its vended input-group and textarea deps are written for React 19 (plain functions with ref-as-prop); this app is on React 18, where those subcomponents drop the refs Base UI passes for focus and anchoring, so InputGroupInput, InputGroupButton, and ComboboxTrigger are adapted to forwardRef. Those ui/ files now diverge from the registry, and a future shadcn add would overwrite the adaptation until the app moves to React 19. Adds class-variance-authority, which input-group needs Give loading skeletons a per-column renderSkeleton escape hatch on the shared DataTable and mirror the Key cell exactly (alias line, secret, status pill), so skeleton rows match the real rows instead of being shorter and simpler Resolve the automated review: the toolbar search and the drawer Key Alias filter both mapped to the key-alias query, so the search silently overrode the drawer value while its chip stayed visible. Consolidate to a single alias search in the toolbar (placeholder now "Search by key alias…") and drop the redundant drawer field. Re-add coverage for the Created By column's alias-over-email display Refine the Team and Organization filters: they match on name and id, so the labels read "Team" and "Organization" rather than "... ID", each option shows the name with the id on a muted second line instead of "name (id)", and the active-filter chip shows the friendly name * chore(ui): drop duplicate class-variance-authority, use the repo cva package in input-group * fix(mcp): relay upstream OAuth token and DCR rejections instead of a generic 500 An upstream token endpoint rejection (e.g. Google requiring client_secret even for PKCE web clients) escaped exchange_token_with_server as a raw httpx.HTTPStatusError, which the global exception handler turned into an opaque 500 Internal server error in the create-flow UI. The RFC 6749 section 5.2 error body the IdP sent (error, error_description, error_uri) is now relayed with the upstream's own 400/401 status; rejections outside the section 5.2 contract map to 502 so a broken upstream is not misattributed to the caller. The same relay covers the non-bridge DCR registration arm, and a 200 token response without a usable access_token now answers 502 instead of a KeyError 500. The catch wraps the post call itself because litellm's AsyncHTTPHandler raises MaskedHTTPStatusError at call time, which also made the pre-existing bridge-relay status check unreachable in production. The dashboard's token exchange error message now composes error and error_description so the form shows the IdP's reason * refactor(mcp): drop bridge relay status check made unreachable by the unified relay The try/except around the registration post now relays every upstream 4xx/5xx for both arms, so the bridge_relay status_code check could never fire; removing it addresses the Greptile P2 dead-code finding * fix(mcp): classify get_user_object's wrapped DB outage across the exception chain get_user_object catches every DB failure in a broad except and re-raises a bare ValueError (litellm/proxy/auth/auth_checks.py), so a real outage and a missing user look identical and the original error survives only as __context__. The dcr_bridge admission path keyed its 503-vs-401 decision on the exception type, so a transient outage during a user-subject reload surfaced as a 401 rather than a retryable 503, and the regression test injected a raw ConnectionError, a shape get_user_object never produces, so it passed on a fiction Add PrismaDBExceptionHandler.is_database_service_unavailable_error_in_chain, which walks __cause__/__context__ (bounded and cycle-safe) the PEP 3134 way, and route _raise_503_if_db_unavailable through it. Move the user's object_permission resolution inside the single classified try so an outage there is a 503 too, never an opaque 500. Pin get_user_object's wrapping with a contract test that drives the real function, and drive the reload tests with that same faithful shape so a chain-blind regression fails them * fix: redact async complete streaming response for custom callbacks (#33106) * fix response not being redacted for custom callbacks with streaming enabled * reduce code duplication * add unit test * fix: resolve lint violations in adopted redaction fix * fix: scope streaming response redaction to the opted-out custom logger --------- Co-authored-by: Moritz Müller <moritz.mueller2@tu-dresden.de> * build(ui): bump @tanstack/react-pacer from 0.2.0 to 0.22.1 (#33041) * refactor(ui): standardize debounce waits behind shared DEBOUNCE_WAIT_MS constant * build(ui): bump @tanstack/react-pacer from 0.2.0 to 0.22.1 * fix(ui): address Virtual Keys redesign review nits (#33112) * fix(ui): address Virtual Keys redesign review nits Restore sorting by budget on the merged Spend / Budget column. The column now uses a new DataTableMultiSortHeader whose chevron opens a menu offering Spend and Budget in both directions plus Reset, so the progress-bar cell stays merged while the sort field becomes an explicit choice. Sorting is server-side, so the chosen field id (spend or max_budget, both accepted by /key/list) flows straight through as sort_by Fill the DataTable to its container width when column resizing is on. The table width was pinned to the sum of column widths, so hiding columns left an empty gutter on the right. It now keeps that width as a minimum and stretches to 100% on underflow while still scrolling on overflow, which also covers the same gap in TeamVirtualKeysTable since both share the component Drop the dark background box behind the page-header icon so the Virtual Keys header reads like the Teams header, and pull the 4-line inline filter lambda in SearchSelect out into a named matchesQuery helper Extends the DataTable and VirtualKeysTable tests to cover the new multi-field sort menu (field id maps to sort_by, active indicator, reset) and the fill-to-container width * fix(ui): emphasize the active field in the Spend / Budget sort header The merged Spend / Budget header always read "Spend / Budget" regardless of which field drove the sort, so after picking Budget descending there was no way to tell what was sorted without reopening the menu. The header now builds its label from the sort fields and emphasizes whichever one is active (bold, full-strength text) while muting the other, so the sorted column reads at a glance alongside the direction chevron. Drops the now-redundant title prop since the label is derived from the fields * fix(ui): remove w-full so the keys page content stops overflowing by 32px The virtual keys content wrapper used "w-full mx-4", which sets the width to 100% of the parent and then adds 16px of horizontal margin on each side, so its margin-box came to 100% + 32px and overflowed the scrollable main region by exactly 32px. That surfaced as a horizontal scrollbar along the bottom of the whole content area, under the pagination. A block div is already full-width, so dropping w-full lets mx-4 inset it correctly with no overflow * fix(ui): darken the clickable Key cell on hover so it reads as clickable The Key cell was the click target that opens the key detail, but hovering only faded the chevron in with no change to the cell itself, so there was no cue that the area was clickable. Give the cell a subtle muted background and a pointer cursor on hover. The button spans the full cell (a negative inline margin plus a matching width offset so the hover fill reaches both cell edges while the title stays aligned with the other columns) * fix(openai/responses): clamp max_output_tokens below API minimum (#33098) * fix(openai/responses): clamp max_output_tokens below API minimum Claude Code sends a max_tokens=1 warmup probe when running /model, which the Anthropic Messages -> Responses adapter forwards as max_output_tokens=1. OpenAI's Responses API rejects values below 16, so the probe failed with a 400. Clamp anything below the minimum up to 16 in map_openai_params so all Responses API entrypoints (direct, chat->responses, anthropic->responses) are covered. Co-Authored-By: Ishaan Jaffer <155045088+ishaan-berri@users.noreply.github.com> * refactor(openai/responses): extract _enforce_min_max_output_tokens helper Co-Authored-By: Ishaan Jaffer <155045088+ishaan-berri@users.noreply.github.com> --------- Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> Co-authored-by: Ishaan Jaffer <155045088+ishaan-berri@users.noreply.github.com> * fix(prometheus): read v3 rate limiter remaining values for per-key model gauges (#33119) * fix(ui): drop w-full from page-content wrappers to remove 32px horizontal overflow (#33118) Several dashboard pages wrap their content in a div styled w-full mx-4, so the element's width is 100% of the scrollable main while mx-4 adds 16px of margin on each side. That makes the margin-box 100% + 32px wide, which overflows main by exactly 32px. Because main uses overflow-y-auto its overflow-x computes to auto, so the overflow surfaces as a horizontal scrollbar along the bottom of the whole content area under the pagination The wrapped block is already full width without w-full, so removing that one token keeps the layout and drops the overflow to 0. This is the same fix already applied to the Virtual Keys page in #33112, extended to the remaining pages that share the wrapper: Models + Endpoints, Tag Management, Organizations, Vector Stores, AI Hub, and Logging & Alerts * refactor(ui): migrate straightforward value debounces to react-pacer (#33042) * refactor(ui): standardize debounce waits behind shared DEBOUNCE_WAIT_MS constant * build(ui): bump @tanstack/react-pacer from 0.2.0 to 0.22.1 * refactor(ui): migrate straightforward value debounces to react-pacer * feat(mcp): client-held refresh envelope for the dcr_bridge oauth_delegate flow A dcr_bridge oauth_delegate access envelope is capped at one hour, and until now the mode had no refresh at all: when the envelope expired the client had to re-run the interactive authorization_code flow. This adds a second client-held credential, the refresh envelope, so the client renews on a back channel and only re-authenticates when the refresh envelope expires or the upstream refresh token dies. The refresh envelope is a distinct llm_refresh_ credential that seals only the upstream refresh token (never the access token) bound to the same litellm identity and MCP server as the access envelope, under the same master-key-derived keys, with nothing stored server-side. Both envelopes now carry a signed kind claim ("access" or "refresh") that open() requires to match, so a refresh envelope can never open as an access credential even if its wire prefix is swapped (the prefix is not signed; the claim is). A refresh envelope presented at the MCP tool-call edge is not an access envelope, so admission fails it closed the same way it already fails any non-access bearer. At the token endpoint the authorization_code mint now returns a refresh envelope alongside the access envelope whenever the upstream returned a refresh token, and the refresh_token grant is supported for bridge servers: the client presents its refresh envelope, the endpoint opens it, re-validates the sealed litellm key so a revoked key cannot keep refreshing, unwraps the real upstream refresh token, exchanges it with the upstream IdP, and returns a fresh access envelope. Because the endpoint re-seals a refresh envelope only when the upstream returns a new refresh token, the design mirrors the upstream's own rotation policy rather than reinventing it: with a rotating upstream the client rotates and reuse is detected upstream; with a non-rotating upstream the original refresh envelope stands until its bounded 14-day TTL. Both preconditions and the unwrap run before the exchange, so a rejected refresh never consumes or rotates an upstream token. The pure envelope and credential layers stay side-effect free: mint/open share one signing, size, and kind gate across both envelope kinds, and every failure is a value. Tests cover the refresh round-trip, the kind-claim and server-id bindings, the revoked-key gate, upstream rotation carried through, the unwrap sending the real upstream token upstream, and edge rejection of a refresh envelope; the three security bindings are mutation-checked. Limitation documented in the PR: gateway-enforced refresh rotation with reuse detection would require server-side state, which this zero-custody mode omits by design, so the refresh envelope inherits the upstream's rotation posture plus gateway identity binding and a bounded TTL. * fix(mcp): reject a refresh envelope explicitly at the tool-call edge The live proof showed a refresh envelope presented at the MCP tool-call edge was rejected, but through the generic oauth2 arm ("expected a virtual key starting with sk-") rather than the bridge arm, because the admission routing gate is_bridge_envelope_shaped matched only the access prefix. The rejection was already fail-closed and never forwarded anything upstream, but the path was imprecise and the unit test modelled a route the real router did not take. Match either envelope kind in is_bridge_envelope_shaped so the bridge arm engages for a refresh envelope too, and have resolve_bridge_envelope return BridgeEnvelopeInvalid for it: a refresh envelope is a valid gateway credential but only ever presented back to the token endpoint, never usable to authenticate a tool call. Admission now fails it closed with the bridge arm's own 401 ("Invalid or expired credential"), live-verified, with the upstream never touched. is_bridge_envelope_shaped has a single caller (the admission routing gate), so the change is contained. * fix(mcp): SecretStr the unwrapped refresh token, drop the dead request arg, fail closed on a missing user Three review findings on the refresh path, addressed at the root: _BridgeRefreshReady.upstream_refresh_token was a plain str, the one credential in the envelope/bridge layer that escaped the SecretStr discipline every other one follows (RefreshCredential.refresh_token, UpstreamTokenGrant.access_token, EnvelopeKeys.signing_key). A repr or a traceback capturing a local _BridgeRefreshReady would have logged the raw upstream refresh token. It is now a SecretStr, carried as the SecretStr open_bridge_refresh_envelope already returns and unwrapped only at the point the exchange builds the upstream request body. _prepare_bridge_refresh took a request it never read; on the refresh path identity comes entirely from the sealed envelope, not the HTTP request, so the parameter was dead and misleadingly implied it read from the request the way the authorization_code prepare does. Removed, and the caller updated. _reload_active_user_by_id misclassified a missing user as unresolvable (500). This is the same root cause as the admission user-reload fix: get_user_object raises a bare Exception for a deleted user rather than a ProxyException, so its except-Exception arm must fail closed to no_active_key (which the refresh path maps to invalid_grant) for anything that is not a database-service-unavailable outage, rather than treating a missing user as an opaque gateway fault. Regression tests cover the missing-user and DB-outage classifications directly. * fix(mcp): make the dcr_bridge refresh path fail correctly on outages, dead tokens, and revoked owners Four fixes to the refresh_token grant for dcr_bridge oauth_delegate, surfaced by an adversarial pass over the exchange path Route the user-subject re-validation's outage check through the chain-aware classifier, so a transient DB outage (which get_user_object wraps in a bare ValueError) reports as unavailable (a retryable 503) rather than collapsing to no_active_key and an invalid_grant, matching how admission now handles the same wrapper When the upstream reports its own refresh token as already elapsed (refresh_expires_in non-positive), do not seal it into a full-TTL refresh envelope; return no refresh so the exchange degrades to an access-only response, mirroring how the access grant refuses an already-elapsed access token instead of capping it When the upstream rejects the sealed refresh token with 400 invalid_grant (revoked or expired at the IdP), return an RFC 6749 invalid_grant response so the OAuth client re-runs authorization_code, rather than surfacing the opaque upstream error it cannot act on Gate key-subject renewal on the owner's SCIM state, mirroring admission's _reject_if_admitted_owner_scim_deactivated, so an offboarded user cannot keep refreshing a still-active key; the check fails open on a missing owner or a DB blip so a key that outlives its owner record does not get wrongly revoked Each fix has a mutation-checked regression test * test(proxy): add regression tests for management_endpoints edge cases (#32976) Mutation testing surfaced branches in cost_tracking_settings and common_utils that the suite executed but never asserted on. Pin those behaviors with targeted tests: the returned (model, provider) from _resolve_model_for_cost_lookup for deployments carrying a custom_llm_provider and for deployments missing the litellm_params / model_info keys, plus the exact error-response bodies, the caller-identity lookup arguments, and the member and guard branches in common_utils. * fix(auto-router): correct Responses API tool_choice shape and propagate alias litellm_params (#32974) * fix(anthropic-messages): send bare-string tool_choice to Responses API, propagate router-alias litellm_params The Anthropic /v1/messages -> Responses API adapter always wrapped tool_choice in an object ({"type": "auto"}, {"type": "required"}), but the Responses API's tool_choice schema for these cases is a bare string ("auto"/"required"/"none"). Sending the object shape to an OpenAI-compatible backend (e.g. vLLM) fails Pydantic validation with a 400. The "none" case also fell through to "auto" instead of mapping to "none". Separately, litellm_params configured directly on a router-alias deployment (auto_router/complexity_router, adaptive_router, quality_router, or semantic auto_router) - e.g. cache_control_injection_points, drop_params - were silently dropped for every request through that alias. async_pre_routing_hook swaps `model` from the alias name to the selected tier/route's model before the deployment lookup runs, so the outbound call only ever merged in the tier deployment's own litellm_params, never the alias's. Register non-routing-config litellm_params from the alias deployment and apply them to the request whenever a pre-routing hook substitutes the model. * fix: satisfy ruff-strict-budget UP006 and router coverage checker Use builtin dict[...] generics instead of typing.Dict for the two new annotations introduced in the previous commit, since they pushed UP006 over the codebase ceiling in ruff-strict-budget.json. Add a direct unit test for _register_pre_routing_alias_overrides so the text-based router_code_coverage.py checker sees it exercised by name. * fix(router): replace alias-param denylist with a tight allowlist _PRE_ROUTING_ALIAS_RESERVED_PARAMS excluded router-init-only keys from the alias's litellm_params before forwarding the rest as request kwargs, but GenericLiteLLMParams also holds deployment-management fields (tpm, rpm, weight, tags, max_budget, budget_duration, use_in_pass_through, litellm_credential_name, ...) on the same object. Any of those left off the denylist would get silently forwarded as if they were request kwargs. Replace the denylist with a tight allowlist of exactly the two request-shaping params this feature exists for - drop_params and cache_control_injection_points - so unrelated management fields never reach the outbound call regardless of what else GenericLiteLLMParams grows to hold. * fix(router): re-register adaptive-alias overrides on set_model_list reload set_model_list() unconditionally clears pre_routing_alias_overrides on every call (e.g. /config/reload), but _finalize_adaptive_router_if_configured() skips rebuilding an AdaptiveRouter whose model_name already exists in self.adaptive_routers - so _register_pre_routing_alias_overrides() never ran again for an auto_router/adaptive_router alias after a reload, silently dropping its drop_params/cache_control_injection_points. Build the Deployment unconditionally and re-register its overrides even on the skip-existing-router path; only the (expensive) AdaptiveRouter construction itself stays skipped. * style: ruff format after merging litellm_internal_staging * fix(router): drop the alias-param allowlist, exclude only model Per review discussion: instead of a router.py-local allowlist of exactly which litellm_params an alias (auto_router/complexity_router, adaptive_router, quality_router, semantic auto_router) can forward to the request it routes, _register_pre_routing_alias_overrides now forwards everything except `model` (the alias marker itself, e.g. auto_router/complexity_router, never a real provider model). Router-init-only fields (complexity_router_config, complexity_router_default_model, auto_router_config, auto_router_config_path, auto_router_default_model, auto_router_embedding_model, adaptive_router_config, adaptive_router_default_model, quality_router_config, quality_router_default_model) now flow into request_kwargs unfiltered too. That's safe because litellm.completion()/acompletion() already strips anything in litellm.types.utils.all_litellm_params before building the provider request - added these 10 keys there, alongside the deployment-management fields (tpm, rpm, weight, ...) already listed. Verified live: without that addition, complexity_router_config lands in extra_body and ships raw to the provider; with it, it's stripped. This moves the "which fields aren't real LLM params" list from a router.py-local allowlist to the single existing global list every completion() call already depends on, instead of maintaining two. * refactor(router): look up alias litellm_params on demand instead of caching them _register_pre_routing_alias_overrides cached each alias's litellm_params into self.pre_routing_alias_overrides at deployment-init time, which required keeping that cache in sync with set_model_list() reloads - the exact bug the previous adaptive-router-reload fix was patching around (AdaptiveRouter survives a reload, but the cache didn't always get refreshed to match). Delete the cache and the registration method entirely. async_pre_routing_hook now looks up the alias's own litellm_params directly from self.model_list via self.model_name_to_deployment_indices at request time, the same model_list that's already correctly rebuilt on every set_model_list() call. No second piece of state to invalidate, so the reload staleness bug class isn't possible anymore, and it's less code than before. * fix(mcp): keep out-of-contract upstream error bodies out of client responses The token and DCR relays serve unauthenticated OAuth clients, so only the RFC 6749/7591 error fields may cross the trust boundary. A rejection body outside those contracts (HTML error page, proxy banner, stack trace) is now logged server-side, bounded, and the client response names only the upstream status. Addresses the Veria information-exposure finding * fix(mcp): re-request the sealed scope on a bridge refresh when the client omits it The refresh envelope seals the upstream scope as the scope to re-request (RefreshCredential), but _prepare_bridge_refresh dropped it, unwrapping only the refresh token, and the exchange added scope to the upstream request only from the client's HTTP form. A DCR/MCP client typically omits scope on refresh, so the sealed scope was never sent and a stricter upstream could narrow or drop the renewed token's scope Thread the sealed scope through _BridgeRefreshReady.upstream_scope and fall back to it when the client sends none; a client-supplied scope still wins, which RFC 6749 section 6 bounds to the original grant. The regression test drives a refresh where the client omits scope and asserts the upstream POST carries the sealed scope, mutation-checked against both the drop and the fallback * fix(ui): render the sidebar scrollbar with shadcn ScrollArea (#33124) * fix(ui): render the sidebar scrollbar with shadcn ScrollArea The sidebar navigation scrolled through a native overflow-y-auto container, so the browser drew its default scrollbar. It now scrolls through the shadcn ScrollArea primitive so the thumb matches the rest of the dashboard Switching to ScrollArea surfaced a latent styling gap. The Base UI scroll-area, tabs, and separator primitives rely on data-horizontal and data-vertical Tailwind variants that resolve to [data-orientation="horizontal"] and [data-orientation="vertical"], and those variants ship in shadcn's shared stylesheet. The project never imported it, so the classes matched nothing and the scrollbar collapsed to zero width. This adds shadcn as a devDependency and imports shadcn/tailwind.css, which also repairs the vertical tabs and separator styling. See shadcn-ui/ui#9196 for the upstream tracking issue * refactor(ui): inline the Base UI data-* variants, drop the shadcn dep The earlier fix imported shadcn/tailwind.css through the shadcn devDependency, which pulled 219 packages and tied the CSS build to shadcn's package exports (an open Turbopack-breaking bug, shadcn-ui/ui#10931). shadcn's model is that we own the components, so the custom variants those components depend on belong in our own stylesheet rather than a runtime dependency. This inlines the nine data-* custom variants and the no-scrollbar utility that the Base UI primitives reference into globals.css, and removes the shadcn package. * refactor(ui): migrate callback debounce sites to react-pacer with regression tests (#33043) * refactor(ui): standardize debounce waits behind shared DEBOUNCE_WAIT_MS constant * build(ui): bump @tanstack/react-pacer from 0.2.0 to 0.22.1 * refactor(ui): migrate straightforward value debounces to react-pacer * refactor(ui): migrate callback debounce sites to react-pacer with regression tests * chore(ui): restore trailing newline in eslint-suppressions.json * test(ui): mock all pacer debounce hooks in VirtualKeysTable test * fix(ui): update merged debounce tests for OldTeams to Teams rename * fix(mcp): carry the requested scope forward when the upstream omits it on a bridge refresh The prior fix sent the sealed scope on a refresh, but the re-minted refresh envelope re-seals scope from the upstream response, and RFC 6749 section 5.1 lets an upstream omit scope when it is unchanged. So after one refresh whose response omitted scope, the new envelope sealed scope=None and every subsequent refresh dropped it, letting a stricter upstream narrow the renewed token When the upstream omits scope on a bridge refresh, seal the scope we requested (which RFC 6749 section 5.1 defines as the granted scope when omitted) into the renewed access and refresh envelopes, so the scope survives the whole refresh chain. The regression test refreshes against an upstream that omits scope, asserts the new refresh envelope still carries it, and refreshes again off that envelope to prove the chain does not lose it, mutation-checked * refactor(mcp): classify upstream OAuth faults once and derive status, code, and prose from the value Replaces the accreted relay helpers with a faults package (types, classify, render_oauth): every upstream token/DCR rejection is classified into exactly one fault value and the response status, wire error code, and prose are all derived from that value, so a caller-fault code can never ship on a server-fault status (the bugbot finding on invalid_grant over a 500). Classification takes the credential source into account: invalid_client and friends against the server's stored credentials are the operator's fault and render as 502 server_error with gateway-authored prose while the IdP's prose stays in server logs; the same codes against caller-supplied credentials relay on the status the code implies. Classifiers are total, so an unreadable rejection body (lying content-encoding, unconsumed stream) yields the same 502 fault instead of resurrecting the opaque 500 (the second bugbot finding); DCR rejections normalize to 400 per RFC 7591 regardless of the upstream's status * style(mcp): unquote annotations and use PEP 604 unions in the faults package * fix(mcp): detect upstream invalid_grant by the RFC 6749 error field, not a body substring The bridge refresh path decided whether an upstream token-endpoint rejection was invalid_grant by substring-matching the raw response body, so a rejection whose actual error is something else but whose error_description merely contains the string invalid_grant would false-match, map to invalid_grant, and trigger a needless authorization_code re-run Parse the RFC 6749 section 5.2 error object and compare the error field. A non-JSON body, or an error that is not invalid_grant, now propagates as the upstream error rather than being reinterpreted. The regression test drives an invalid_client rejection whose description contains the string invalid_grant and asserts it is not mapped, mutation-checked against the substring match * fix(mcp): keep upstream self-blame codes and gateway capability gaps off the caller Extends the fault matrix per review: server_error and temporarily_unavailable are codes by which the upstream blames itself, so they classify as a new UpstreamReportedFault arm rendering 502/503 with a matching wire code instead of a 400 that blames the caller; invalid_target is a gateway capability gap (RFC 8707 resource indicators, LIT-4339) and is gateway-blamed regardless of whose credentials were presented; the DCR classifier shares the same blame assignment. The gateway-fault arm is renamed GatewayRejected since it now covers capability gaps as well as stored-credential rejections * chore: add CODEOWNERS for ui and proxy UI build artifacts (#33131) * feat(ui): rebuild the Teams table on the shared DataTable (#33128) * feat(ui): rebuild the Teams table on the shared DataTable The Your Teams tab moves off the Ant Design table onto the shared DataTable that the Virtual Keys page uses, following the new dashboard design. It gains server-side sort, pagination and filtering, a toolbar with a filter drawer and a columns menu, and a per-row actions menu Sorting is wired only to the columns /v2/team/list can actually order by (team_alias, created_at); Spend / Budget and Updated stay unsorted because the endpoint silently ignores those fields. The design's "Created by" column is dropped since the team object has no such field, and the drawer's "Has keys" filter is dropped for the same reason. The Resources cell shows members, models and keys as colored pills, and the actions menu keeps the existing Edit, Copy team ID and Delete behaviors, with Edit and Delete gated to Admin The teams grid gets its own unit tests in TeamsPage/TeamsTable.test.tsx. Teams.tsx keeps the create-team modal, delete modal, detail view and tabs, now refreshing the list through React Query invalidation instead of a manual refetch * fix(ui): match Teams loading skeletons to the rendered row height The default twoLine and chips skeleton shapes rendered the Team and Resources cells shorter than the loaded row (a real row measures 55px, the old skeleton ~49px), so the loading state looked visibly squat. Give the Team column a custom renderSkeleton that mirrors the two-line IdentityCell (measured 54px) and the Resources column one that mirrors the pills, and mark the hidden Rate Limits column as two-line so it matches when shown * fix(ui): keep team admins' Members tab by deriving is_team_admin from the selected team The redesign computed is_team_admin from useTeam(selectedTeamId), but that hook returns teamInfoCall's nested { team_info: { members_with_roles } } shape, so the top-level members_with_roles read was always undefined and is_team_admin was always false. For a non-proxy-admin team admin that hid the Members, Member Permissions and Settings tabs in the team detail view, which broke the team-admin add/remove member e2e tests. Pass the Team object up from the table instead (/v2/team/list returns it with a top-level members_with_roles), matching the pre-redesign behavior; proxy admins were unaffected because is_proxy_admin already granted access Also point the Delete-a-team e2e at the new kebab: open the row actions menu, then click Delete team, rather than clicking the old inline delete icon * fix(keys): persist key_type so the UI shows correct key scope instead of "All Proxy Models" (#33115) * fix(ui): derive key model scope so SCIM/management/read-only keys stop showing 'All Proxy Models' key_type is not persisted on a key (the proxy maps it to allowed_routes and drops it), so the keys tables only inspected the models list and rendered 'All Proxy Models' for any key with an empty models array, including SCIM, Management and Read-only keys that cannot call a single model. Add deriveKeyModelScope(allowed_routes) and render 'No model access' with a scope tooltip for those recognized scopes; unrestricted, AI-API and custom keys keep the existing model-list rendering. Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * refactor(ui): move key_scope helper to components root Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * feat(keys): persist key_type on virtual keys so the UI reads scope directly Add a nullable key_type column to LiteLLM_VerificationToken (root, proxy, and proxy-extras schemas plus an additive migration) and stop dropping the value in handle_key_type, so management/read_only/llm_api/default keys store their type alongside the derived allowed_routes. Surface it on the key read and create response models. The dashboard now prefers the persisted key_type for the no-inference buckets and keeps the allowed_routes derivation as the fallback for keys created before the column existed (key_type null), so no backfill is required. Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * style(keys): use PEP604 X | None for new key_type annotations to satisfy ruff UP045 budget Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(keys): add key_type column to LiteLLM_DeletedVerificationToken The deleted-token archive model inherits key_type from the verification token, so regenerate/delete flows write key_type into LiteLLM_DeletedVerificationToken. Add the column (all schemas + migration) so the archive insert does not fail with FieldNotFoundError. Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(migrations): regenerate key_type migration via runbook (canonical ADD COLUMN) Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> --------- Co-authored-by: ryan <ryan@berri.ai> Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * feat(router): opt-in session affinity for complexity router (#33126) * feat(router): opt-in session affinity for complexity router Complexity router reclassified every turn, which could flip the routed model group mid-session and break provider-side prompt caching. Add a session_affinity config flag: when a session_id is resolvable, pin the model chosen on the first turn and reuse it for the rest of the session, skipping reclassification. Pinned turns still stamp the adaptive bandit's chosen-model metadata so reward feedback keeps working when adaptive=True. * fix(router): refresh session-affinity TTL on hit, scope pin by API key Two issues from review: the TTL was only set on the first classification, so an active session outliving session_affinity_ttl_seconds silently lost its pin instead of refreshing as documented. And the cache key was scoped only by session_id, which is client-supplied and unauthenticated, so two different callers reusing the same session_id could poison each other's routing pin. Refresh the TTL on every cache hit, and namespace the cache key by the proxy-derived API key hash when available. * feat(prometheus): expose video duration and image count consumption metrics (#33138) * test(e2e): otel trace completeness on /chat/completions (#33132) * test(e2e): OTEL trace completeness on /chat/completions against a local Jaeger destination Adds the logging-suite infrastructure for LIT-3787 trace-completeness coverage: a jaeger service in the compose stack as the OTEL v2 destination (arize_phoenix preset pointed at it via PHOENIX_COLLECTOR_HTTP_ENDPOINT, so gen-AI spans export through a preset-owned provider - the code path where trace splits happen), a typed Jaeger query read-back client, and the first test: one successful non-streaming /chat/completions call exports ONE complete trace (root SERVER span + auth/db/cost children + gen-AI CLIENT span, no dangling parents). * test(e2e): harden the otel trace read-back per review Jaeger reads now query server-side by the litellm.call_id span tag instead of paging recent traces and filtering client-side; the compose stack's background jobs alone can push a request trace past the page. A failed query hard-fails instead of reading as an empty result, the settle predicate now also waits for the prefix-matched db span the assertion demands, parent-chain walking follows CHILD_OF references only, the zero-trace and split-trace failures get distinct messages, jaeger gets a healthcheck so the depends_on condition is accurate, and the chat docstring names the route the code actually asserts * test(e2e): author the chat trace test docstring * Update logging section in CLAUDE.md Removed mention of OTEL trace-tree completeness from logging integration section. * fix(sso): paginate through all pages when fetching service principal group assignments (#33149) get_group_ids_from_service_principal only read the first page of the Graph API appRoleAssignedTo response, so tenants with more than 100 groups assigned to the enterprise application silently lost group memberships during SSO login. Loop over @odata.nextLink with the same MAX_GRAPH_API_PAGES cap that get_user_groups_from_graph_api already uses, and warn when the cap is hit. Ported from #32792 by @saisurya237 so CI can run. Fixes #32790 Co-authored-by: saisurya237 <saisurya.abhishek237@gmail.com> * test(e2e): otel trace completeness on /v1/messages (#33133) * test(e2e): OTEL trace completeness on /v1/messages Extends the LIT-3787 trace-completeness suite to the Anthropic-native route: one successful non-streaming /v1/messages call must land at the destination as ONE connected trace (root SERVER span + auth/db/cost children + gen-AI CLIENT span, no dangling parents). Adds the raw /v1/messages sender to the logging suite client. * test(e2e): reuse the shared AnthropicMessagesBody per review Drops the duplicate /v1/messages request model in favor of the one models.py already provides (budget_client uses the same one), passes max_tokens at the call site to match the sibling chat test, notes in the docstring why the gen-AI span is named chat on this surface, and adopts the hardened read-back signature * test(e2e): author the messages trace test docstring * test(e2e): declare the messages surface on the covers marker * test(e2e): otel trace completeness on /v1/responses (#33134) * test(e2e): OTEL trace completeness on /v1/responses Extends the LIT-3787 trace-completeness suite to the OpenAI Responses API route: one successful non-streaming /v1/responses call must land at the destination as ONE connected trace. Adds the raw /v1/responses sender, a CHEAP_OPENAI_MODEL config constant, and registers responses in the otel registry cell's exercised_on. * test(e2e): author the responses trace test docstring * test(e2e): declare the responses and chat surfaces on the covers markers * feat(ui): add adaptive routing settings to Auto-Router v2 (#33146) * refactor(mcp): extract the dcr_bridge token flow into bridge_token_flow.py discoverable_endpoints.py had grown to 2695 lines mixing FastAPI route handlers with the dcr_bridge token-flow logic, against the no-monster-files convention. This moves the bridge token flow (the litellm-key/user resolution, the SCIM revalidation gate, and the mint/refresh envelope logic with their types and error mappers) into a dedicated bridge_token_flow.py, leaving the route handlers and the shared exchange_token_with_server orchestrator in discoverable_endpoints.py importing from it Pure relocation, zero behavior change. The moved code is byte-verbatim except one type annotation quoted as a forward reference (_BridgeAuthorizationCode is used only for typing and imported under TYPE_CHECKING to avoid a cycle), and the new module imports nothing from discoverable_endpoints at runtime. 275 tests pass unchanged; the test patch targets for moved internals were repointed to the new module and verified to still apply * bump: litellm-enterprise 0.1.49 -> 0.1.50, litellm-proxy-extras 0.4.76 -> 0.4.77, litellm 1.93.0 -> 1.94.0 (#33229) * chore(deps): pin httplib2 and setuptools transitive floors (#33233) Raise the constraint floors for two transitive dependencies so resolution moves them to their latest maintenance releases: httplib2 0.31.2 -> 0.32.0 and setuptools 82.0.1 -> 83.0.0. Both are pulled in only by optional integrations (Google API client, grpc tooling, lunary observability, the nvidia-riva extra), all lower-bound only, so the floors stay inside every requirer's allowed range and a default install is unaffected * feat(ui): left-anchor the Create Key and Create Team CTAs (#33248) Move the Create New Key and Create Team buttons out of the page header's right-side action slot. On Teams the button now sits in the tab bar's left slot, separated from the three tabs by a vertical rule, so the CTA and tabs read as one left-anchored cluster. On Keys, which has no tabs, the button anchors left on its own row beneath the title. * fix(anthropic/passthrough): drop incompatible temperature when downgrading adaptive thinking for pre-4.6 models (#33244) * fix(anthropic/passthrough): drop temperature and cap thinking budget when downgrading adaptive thinking for pre-4.6 models * test(anthropic/passthrough): use sufficient max_tokens for reasoning_effort thinking mapping * fix(anthropic/passthrough): drop incompatible temperature when downgrading adaptive thinking for pre-4.6 models Narrow the fix to the temperature reconciliation; the reasoning_effort budget cap is reverted because the live translation grid relies on budget_tokens >= max_tokens to reject unsupported effort tiers (xhigh/max) on budget-mode models, so capping turned those 400s into 200s. --------- Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(guardrails): run apply_guardrail-style model-level pre_call guardrails at deployment hook (#33136) * fix(guardrails): run apply_guardrail-style model-level pre_call guardrails at deployment hook * fix(guardrails): keep request-body dispatch predicate unchanged * fix(guardrails): fail closed when proxy extras are missing at deployment hook * fix(proxy)!: enforce user budget on team keys (read-time + reservation) with UI opt-out (#32005) * fix: enforce user budget on team keys User budget was skipped when the key belonged to a team, letting users exceed their personal budget by going through a team key. Remove the team_object guard in _user_max_budget_check so user budgets are always enforced. Add skip_user_budget_on_team_key general_settings flag to opt back into the old behavior. Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix: update test to expect user budget enforcement on team keys Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(proxy): enforce user budget on team keys in reservation path and expose skip flag in UI Extends the read-time fix so the optimistic budget reservation also reserves the user spend counter for team-scoped keys, register skip_user_budget_on_team_key in ConfigGeneralSettings so /config/field/update accepts it, and surface it as a Boolean toggle on the Admin UI General Settings table via allowed_args in /config/list. Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * test: assert budget_exceeded ProxyException in personal budget test Tighten the broad pytest.raises(Exception) so the test only passes when the auth flow rejects with a budget_exceeded ProxyException, and switch the new ConfigGeneralSettings field to Optional[bool] to match the surrounding annotation style * fix: revert to bool | None to stay under UP045 strict budget --------- Co-authored-by: Krrish Dholakia <krrishdholakia@berri.ai> Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> Co-authored-by: ryan-crabbe-berri <ryan@berri.ai> * fix(e2e): bound spend-log snapshots to a /spend/logs/v2 window (#33265) The rate-limited batch spend test snapshotted unattributed rows via the unpaginated /spend/logs whole-table read, which grows with the environment (58MB on stage) and OOMKilled the e2e runner at its 512Mi limit on every scheduled run. Gateway.spend_logs_window pages /spend/logs/v2 over an explicit date window instead, and SpendLogsParams now rejects a filterless read so the whole-table call cannot come back * refactor: make the code easier to read * feat(pricing): add gemini-omni-flash-preview with video output token pricing * fix(gemini): map video response modality instead of MODALITY_UNSPECIFIED * fix(anthropic): use native output capability (#33235) * fix(anthropic): route native structured output Use model capability metadata so new native structured-output models do not require transformation allowlist changes. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(anthropic): pass provider to capability Co-authored-by: Cursor <cursoragent@cursor.com> * test(anthropic): cover dotted model IDs Co-authored-by: Cursor <cursoragent@cursor.com> * fix(anthropic): handle remote capability lag Co-authored-by: Cursor <cursoragent@cursor.com> --------- Co-authored-by: Cursor <cursoragent@cursor.com> * fix(ci): retry setup-uv installs to survive transient manifest fetch failures * docs(e2e): add cache_hit to the naming grammar assertion vocabulary * refactor(e2e): share anthropic cache-control shapes in endpoints_client * fix(proxy): never log raw virtual keys in key insertion debug output (#33268) * fix(proxy): never log raw virtual keys in key insertion debug output * fix(proxy): tolerate None token in insert_data debug log redaction * fix(auth): scope the JWT enterprise gate to actual JWTs (#33296) With enable_jwt_auth enabled but no enterprise license (premium_user False), the JWT premium check fired on every request before the token was inspected, so the master key, sk- virtual keys, and the encrypted CLI/UI SSO session token that `lite login` issues all 401'd with "JWT Auth is an enterprise only feature" and were never decoded. That broke `lite login`, `lite claude`, and the proxy master key on any deployment that turned JWT auth on without a license. Move the premium check inside the is_jwt branch so it gates only real JWTs. Non-JWT credentials fall through to their own auth paths regardless of license; actual JWTs still require premium, so the enterprise gate is unchanged for the feature it protects. * test(e2e): scope virtual keys to the deployment under test * fix(s3): sanitize slashes in response-id-derived object key file name (#33271) * refactor(ui): migrate guardrails table onto shared DataTable (#33303) * feat(ui): migrate guardrails table onto shared DataTable Move the guardrails list onto the shared DataTable + cell library as the proof-of-concept for the simple-tables design migration, following the Teams reference pattern. Split the table into a thin container (guardrail_table.tsx) and column defs (guardrailTableColumns.tsx): client-side sort defaulting to created_at desc, a search + refresh toolbar, IdCell / DateCell / StatusBadge cells, real provider logos, a rich empty state, and skeleton loading rows. Row actions move into a per-row overflow menu; deletion stays disabled for config-file guardrails, now surfaced as a disabled menu item instead of a greyed trash icon. Detail view and the delete modal remain owned by GuardrailsPanel. Restyle the "Add New Guardrail" control to the shared Button + dropdown menu. Update the regression tests for the menu-based actions and drop the now-stale eslint suppression entry that the rewrite eliminated. * fix(ui): match guardrails table to the design Address design-review feedback on the guardrails migration: - Drop the search + refresh toolbar. The original table had neither and the SimpleTable design has no toolbar; the container now just renders the sorted table and its empty state. - Give the Guardrail ID cell the design's hover affordance by rendering it with the shared IdentityCell (monospace, chevron on hover) instead of the blue IdCell pill. - Stop pinning the actions column. Pinning added a sticky divider that the design and the Teams table don't have; it is now a plain right-aligned menu column, matching Teams. * fix(ui): match loading skeleton row height to loaded rows The compact skeleton row did not carry the h-8 height that real compact rows get, so loading rows rendered shorter than loaded ones and the table height jumped when data arrived. Mirror the same size-based height on the skeleton row in the shared DataTable so every compact table loads at a stable height * test(ui): drop stale onGuardrailUpdated from guardrails table baseProps The prop was removed from GuardrailTableProps when the toolbar went away; the test baseProps still listed it. Harmless at the call site since it is spread rather than an object literal, but dead and worth removing * fix(ui): remove dead edit_guardrail_form after guardrails migration The guardrails table migration dropped the last import of EditGuardrailForm, which knip flags as an unused file. The form was already unreachable before the migration: the table wired a delete button only, and nothing ever called handleEditClick to open the modal, so the import was the sole thing keeping the file referenced. Delete it and prune its now-stale eslint suppression entry. Guardrail editing is unchanged and lives in the detail view (GuardrailInfoView) * feat(guardrails): streaming text transformation in generic_guardrail_api (#33110) * feat(guardrails): support streaming text transformation in generic_guardrail_api * chore(guardrails): address PR review feedback * fix(guardrails): fail closed on tool-call and prefix-rewrite leaks in streaming transform * fix(guardrails): address Bugbot review on streaming transform correctness * fix(guardrails): coerce holdback in handler for in-process guardrails * fix(guardrails): harden streaming transform (holdback coercion, tool-call passthrough, n>1 finish_reason) * test(guardrails): targeted _mode_matches coverage for all guardrail_mode shapes * fix(guardrails): inspect streamed tool calls and harden incremental_diff edge cases * test: move ComplianceChecker mode tests to the compliance PR * fix(guardrails): strip content from tool-call passthrough so streamed text can't bypass the transform * fix(guardrails): four correctness fixes for incremental_diff streaming path Four bug fixes on top of the OSS PR's incremental_diff streaming text transformation, all inside the incremental_diff code paths only. No existing block_only, non-streaming, or pre_call behavior is touched. Fix #1 — Mixed content+tool_call finish_reason ordering _tool_call_passthrough_chunk now takes an optional finish_reason_per_choice map. For a choice carrying both delta.content and delta.tool_calls, finish_reason is stripped from the passthrough and recorded on the map so the final synthetic text chunk delivers it. Without this, SSE-compliant clients stopping at finish_reason drop the guardrailed text — defeating the redaction the whole feature exists for. (Greptile P1 twice, Veria.) Fix #2 — Choice index sort in _process_streaming_transform indices/texts_to_check were derived from dict insertion order. For n>1 streams where choice 1 emits before choice 0, guardrail-returned texts aligned to the input order mapped back to the wrong choice indices on write-back — wrong text goes to wrong choice. Sort raw_by_index.keys() up front so realignment is deterministic. (Bugbot Medium.) Fix #3 — Cross-chunk pre-tool-call text flush With default streaming_sampling_rate=5, text chunks followed by a pure tool-call chunk carrying finish_reason='tool_calls' would emit the passthrough with finish_reason before any transformed text delta had fired. Same failure mode as fix #1 but cross-chunk. Now we flush any accumulated text via _round(is_final=False) BEFORE yielding the tool-call passthrough. (Greptile P1.) Fix #4 — Terminator chunk for deferred finish_reason on empty mutated_text _build_transform_chunk returned None early when mutated_text_per_choice was empty. If a mixed content+tool_call chunk had deferred its finish_reason (via fix #1) and the guardrail then suppressed the text (empty return), the deferred finish_reason was never delivered. Now on is_final=True with empty mutated_text_per_choice, we emit a terminator carrying finish_reason per choice from finish_reason_per_choice. (Bugbot High.) Also normalized Optional[X] → X | None across the OSS PR's added surface via ruff UP045 autofix to keep the strict-rule gate within budget. Pure mechanical typing style change, no semantic effect. Regression tests for all four fixes: - test_mixed_chunk_finish_reason_arrives_after_transformed_text (#1) - test_text_flush_precedes_tool_call_passthrough (#3) - test_final_finish_reason_flushed_when_guardrail_suppresses_text (#4) - test_transform_sends_texts_sorted_by_choice_index (#2) All fixes reachable only when streaming_transform_mode == 'incremental_diff' is configured (via _run_incremental_transform_stream) or when a StreamTransformSink is present (via _process_streaming_transform). Verified scope-clean: no changes to block_only, non-streaming, pre_call, moderation, or sibling guardrails. --------- Co-authored-by: Marton Schneider <marton@schneider.co.nl> * test(claude_code): move the Claude Code compatibility matrix under tests/e2e (#32548) * test(claude_code): move the Claude Code compatibility matrix under tests/e2e Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * ci(claude_code): drop the CircleCI compat PR gate; the matrix runs in the scheduled e2e suite instead Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * ci: restore the upload-coverage job dropped by mistake with the compat gate Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(e2e/claude_code): print rate-limit summary on failed compat runs and fix stale run_daily.sh header comments * test(claude_code): assert fine-grained tool streaming via input_json_delta instead of an event-count floor Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> --------- Co-authored-by: mateo <mateo@berri.ai> Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> --------- Co-authored-by: Krrish Dholakia <krrish+github@berri.ai> Co-authored-by: Mateo Wang <277851410+mateo-berri@users.noreply.github.com> Co-authored-by: Deepanshu <deepanshu.lulla@alpha-sense.com> Co-authored-by: yucheng-berri <yucheng@berri.ai> Co-authored-by: Abhimanyu Kapur <38531241+akapur99@users.noreply.github.com> Co-authored-by: tin-berri <tin@berri.ai> Co-authored-by: ryan-crabbe-berri <ryan@berri.ai> Co-authored-by: mubashir1osmani <mubashir.osmani777@gmail.com> Co-authored-by: Shivam Rawat <shivamrawat@Shivams-MacBook-Pro.local> Co-authored-by: Cursor <cursoragent@cursor.com> Co-authored-by: Shivam Rawat <shivam@berri.ai> Co-authored-by: Claude Fable 5 <noreply@anthropic.com> Co-authored-by: Thibault Serot <thibault@linktr.ee> Co-authored-by: Sameer Kankute <sameer@berri.ai> Co-authored-by: devin-ai-integration[bot] <158243242+devin-ai-integration[bot]@users.noreply.github.com> Co-authored-by: Mubashir Osmani <mubashir@berri.ai> Co-authored-by: Or Gershoni <org@lasso.security> Co-authored-by: Moritz Müller <moritz.mueller2@tu-dresden.de> Co-authored-by: Ishaan Jaffer <155045088+ishaan-berri@users.noreply.github.com> Co-authored-by: saisurya237 <saisurya.abhishek237@gmail.com> Co-authored-by: Krrish Dholakia <krrishdholakia@berri.ai> Co-authored-by: Marton Schneider <marton@schneider.co.nl> Co-authored-by: mateo <mateo@berri.ai>	2026-07-14 19:30:42 -07:00
yucheng-berri	a545c493d7	fix(otel): hashable scope for _emit_once when guardrail_mode is list (#31262) ... * fix(otel): hashable scope for _emit_once when guardrail_mode is list `_emit_once` keys `spans_logged` by `(class, id, *scope)`. When a guardrail entry's `guardrail_mode` arrives as a `List[GuardrailEventHooks]` (the shape Presidio expands to with `output_parse_pii: true`, and the shape `event_hook` carries for any `mode: [...]` in config), the tuple contains a list and `spans_logged.get(dedupe_key)` raises `TypeError: unhashable type: 'list'`. On the post-call path this fires inside the logging callback and is swallowed; the request returns 200 but the OTEL `guardrail` span is silently dropped. On the blocking path the same error surfaces as HTTP 500. Adds `_freeze_for_dedupe`, a small recursive normalizer that turns lists and tuples into tuples, sets into frozensets, dicts into frozensets of `(key, value)` pairs, and falls back to `repr` for arbitrary unhashables. Applied inside `_emit_once` before the dict lookup, so all three callsites are protected without touching the guardrail-specific callsite. Helper assumes acyclic input; `guardrail_mode` values are built fresh from config (str enums, lists of str enums, TypedDict of str/list-of-str), so no cycle can arise in practice. Regression tests in `TestOpenTelemetrySpanDedupe` cover the list crash, distinct-list-scope collision, dict and set scope parts, and an end-to-end `_create_guardrail_span` exercise that confirms exactly one `guardrail` span is emitted across repeated lifecycle entrypoints. Each new test fails on a reverted helper (4/4 mutation kill) * fix(otel): cap _freeze_for_dedupe recursion depth and ignore in recursive detector CI's recursive_detector blocks new recursive functions in litellm/ unless they are in the allowlist with a documented bound. Cap the helper at 16 levels and return repr(value) past the cap; this is well past the realistic depth of guardrail_mode (1-3 levels) and means a future caller passing a cyclic container can no longer push the proxy logging path into a RecursionError. Add a regression test that exercises the cycle path. * refactor(otel): annotate _freeze_for_dedupe return as a HashableScope union Per review feedback from @mateo-berri: replace the loose `-> object` annotation with a recursive `HashableScope` union (str | int | float | bool | bytes | None | Tuple[HashableScope, ...] | FrozenSet[HashableScope]) so the helper's contract is visible at the signature. Replace the `try/except hash(value); return value` passthrough with an explicit isinstance check over the hashable-scalar types so the type checker can narrow without requiring `cast(Hashable, value)` on the return. Symmetric: dict keys also flow through the freezer (a TypedDict key is already a string in practice, so behaviorally identical). All 16 regression tests still pass; mutation kill behavior preserved * fix: avoid explicit casting --------- Co-authored-by: Mateo Wang <277851410+mateo-berri@users.noreply.github.com>	2026-06-25 11:59:35 -07:00
yucheng-berri	c2e06890ad	fix: tighten role-based visibility of config and MCP fields (#30587) ... * fix: redact config and MCP secrets in read-only admin views GET /config/field/info and the MCP server list/detail endpoints returned secret-bearing fields to any caller with an admin view, including read-only admins. They now return those fields in full only to a full PROXY_ADMIN; every other caller gets the reduced, non-admin view, while non-sensitive fields remain readable. Regression tests cover the role-based visibility on both endpoints, including that a full admin still sees everything needed to populate the edit form. * fix: redact nested secrets in config field info for non-admins /config/field/info returned structured general_settings fields verbatim to any admin-view caller, so a view-only admin reading database_args received the nested aws_web_identity_token (a DynamoDB role-assumption credential) in plaintext. Recurse into dict/list field values and redact secret leaves for non-PROXY_ADMIN callers, leaving non-secret siblings and full-admin reads unchanged * fix: redact secret config values in /config/list for non-admins /config/list shared the same _user_has_admin_view gate as /config/field/info but returned each field value unredacted, so a view-only admin reading the list received pass_through_endpoints upstream Authorization headers verbatim. Route every general_settings value through a shared role-aware redactor (extracted from /config/field/info) covering the top-level and nested field paths, so non-PROXY_ADMIN callers get secret-bearing fields redacted while full-admin reads stay unchanged * chore(ci): allowlist _redact_secret_values_in_obj in recursive_detector The config secret redactor recurses over JsonValue, which is acyclic, and its depth is bounded by the operator-authored general_settings schema. Add it to the recursive_detector ignore list alongside the other bounded nested-redaction helpers (mask_dict, _redact_sensitive_litellm_params) * proxy: cap recursive secret redaction depth at 10 Match the cap on _redact_sensitive_litellm_params (the closest analog in the proxy, also recursive, key-name driven, returns a sentinel). The previous justification — bounded by operator-authored schema depth, JsonValue acyclic — is true today but is a property of the threat model, not an enforced invariant of the function. If a code path is ever added that pipes external input into general_settings (config import, migration tooling, JWT-driven settings, …) the assumption silently breaks. A local cap makes the invariant local. The cap branch fails closed: at _REDACT_SECRET_MAX_DEPTH the whole subtree is replaced with 'REDACTED' rather than returned verbatim. A future refactor that flips this to fail-open would let a deeply nested credential leak; the new regression test test_redact_secret_values_in_obj_fails_closed_at_max_depth guards against that. Updates the recursive_detector ignore-list rationale to point at the numeric cap rather than the structural argument. * test: actually exercise the depth cap in fails-closed test The previous fixture stored the leaf under the secret-named key 'aws_web_identity_token', which the recursor's key-name short-circuit redacts regardless of the cap — so the test passed both with and without the cap in place. Empirically confirmed: under an uncapped mutant the old fixture still hides the secret (key-name catches it), the new fixture leaks it (only the cap can stop it). Swap the leaf key to a non-secret name so the cap is the only redaction path exercised, making the test fail on mutation as advertised.	2026-06-23 15:25:30 -07:00
Mateo Wang	492891cad8	CI: copy of #25177 (OCI GenAI: embeddings, streaming/reasoning fixes, model catalog) (#28223) ... * fix(opentelemetry): JSON-serialize dict metadata fields for OTEL span attributes (#27451) (#27455) Squash-merged by litellm-agent from Anai-Guo's PR. * feat(dashscope): add embeddings and reranks(qwen3-rerank) support via OpenAI-compatible endpoint (#27508) Squash-merged by litellm-agent from yimao's PR. * fix(vertex_ai/gemini): raise BadRequestError when image_url or url fi… (#24550) Squash-merged by litellm-agent from krisxia0506's PR. * fix(vertex_ai): raise error on mid-stream 429/error chunks instead of silently swallowing (#23711) Squash-merged by litellm-agent from krisxia0506's PR. * fix: raise BadRequestError for file content blocks missing 'file' sub… (#24503) Squash-merged by litellm-agent from krisxia0506's PR. * Fix Gemini MIME detection for extensionless GCS URIs (#27278) Squash-merged by litellm-agent from krisxia0506's PR. * fix(vertex_ai/partner_models): drop unused vertexai SDK gate from count_tokens (closes #28084) (#28107) Squash-merged by litellm-agent from voidborne-d's PR. * feat(chart): add support for autoscaling behavior in HPA (#27990) Squash-merged by litellm-agent from FabrizioCafolla's PR. * feat(proxy): add blocked flag to models for pause/resume from the UI (#27927) Squash-merged by litellm-agent from Cyberfilo's PR. * fix: pass socket timeouts to Redis cluster clients (#27920) Squash-merged by litellm-agent from tomdee's PR. * Fix/cache token (#28009) Squash-merged by litellm-agent from escon1004's PR. * fix(deepseek): forward reasoning_content in multi-turn thinking mode conversations (#28080) Squash-merged by litellm-agent from Divyansh8321's PR. * fix(guardrails): return HTTP 400 instead of 500 for blocked requests (#27617) * fix: reset org and tag budgets (#27326) * reset org budgets * reset tag budgets --------- Co-authored-by: Michael Riad Zaky <michaelr@Mac.localdomain> * fix(ui): omit allowed_routes from key edit save when unchanged (#27553) * fix(ui): omit allowed_routes from key edit save when unchanged When a team admin opens Edit Settings on a key with key_type=AI APIs and saves without changing anything, the UI re-sends the existing allowed_routes value, which the backend's _check_allowed_routes_caller_permission gate rejects for non-proxy-admins (LIT-2681). Strip allowed_routes from the patch in handleSubmit when it deep-equals the original keyData.allowed_routes. The backend treats absence as "leave alone," so no-op saves now succeed for non-admins. Admins explicitly editing the field still send the new value. * fix(ui): order-insensitive allowed_routes diff + cover null-original case Address Greptile review: - Switch the "is allowed_routes unchanged" check to a Set-based comparison so a server-side reorder of the array doesn't register as a user edit and re-trigger LIT-2681. - Add two regression tests: (1) keyData.allowed_routes is null and the form is untouched — patch should strip the field; (2) server returned routes in a different order than the user originally entered — patch should still recognize the value as unchanged. * chore(ui): strip ticket refs and tighten comments in key edit fix - Remove internal-tracker references from in-code comments - Tighten the WHY comment in handleSubmit to two lines - Drop redundant test-block comments — test names already describe the case * fix(ui): annotate Set<string> generic in allowed_routes diff to fix tsc * fix(guardrails): return HTTP 400 instead of 500 for guardrail-blocked requests GuardrailRaisedException and BlockedPiiEntityError both lacked a status_code attribute. When these exceptions reached the proxy exception handler (getattr(e, 'status_code', 500)), the fallback defaulted to HTTP 500 — making intentional guardrail blocks indistinguishable from server errors and causing unnecessary client retries. Changes: - Add status_code=400 (keyword-only) to GuardrailRaisedException - Add status_code=400 (keyword-only) to BlockedPiiEntityError - Update _is_guardrail_intervention() to recognize both exceptions so downstream loggers record 'guardrail_intervened' instead of 'guardrail_failed_to_respond' - Add 6 unit tests for default/custom status codes and getattr pattern - Strengthen existing blocked-action test with status_code assertion Fixes #24348 --------- Co-authored-by: Michael-RZ-Berri <michael@berri.ai> Co-authored-by: Michael Riad Zaky <michaelr@Mac.localdomain> Co-authored-by: ryan-crabbe-berri <ryan@berri.ai> Co-authored-by: Krrish Dholakia <krrish+github@berri.ai> * fix(router/proxy): address Greptile P1+P2 review comments on PR #28161 - router: raise ServiceUnavailableError (503) instead of RouterRateLimitErrorBasic (429) when a specifically-addressed deployment is administratively blocked; 429 misleads retry-enabled clients into spinning forever against a paused model - proxy_server: compute get_fully_blocked_model_names() once before both branches in model_list() instead of duplicating the call in each branch - deepseek: upgrade silent debug log to warning when injecting placeholder reasoning_content so callers are clearly notified of degraded multi-turn quality - tests: update two blocked-deployment assertions to expect ServiceUnavailableError Co-authored-by: Cursor <cursoragent@cursor.com> * fix: address bug detection findings (cache token order, mutable defaults) Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix: address bugs in async pass-through, anthropic cache token detection, rerank tests - async_get_available_deployment_for_pass_through: enforce blocked check on specific deployments - cost_calculator: detect anthropic-style usage by attribute presence (not truthiness) to avoid mixing OpenAI cached_tokens into anthropic normalization when read=0 - dashscope rerank tests: pass request to httpx.Response constructions for consistency Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix code qa * fix(vertex_ai/gemini): strip MIME parameters from GCS contentType GCS object metadata's contentType field can include parameters such as 'text/html; charset=utf-8'. Strip them in _apply_gemini_mime_type_aliases so downstream get_file_extension_from_mime_type sees a bare MIME type. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(vertex_ai/gemini): clarify mime-type error message string concatenation Co-authored-by: Yassin Kortam <yassin@berri.ai> * feat(oci): add embeddings, fix streaming/reasoning, expand model catalog - Add OCIEmbedConfig with full Cohere embed support (7 models, batch up to 96) - Fix sync streaming: split SSE events on \n\n before JSON parsing - Fix reasoning models (Gemini 2.5, xAI Grok): make completionTokens and message optional in OCIResponseChoice to handle max_tokens exhausted on reasoning - Fix compartment_id resolution in chat transform to use resolve_oci_credentials - Fix tool call id: make OCIToolCall.id optional, generate UUID fallback for providers (Google via OCI) that omit it - Add OCI_KEY env var support for inline PEM keys - Fix datetime.utcnow() deprecation in request signing - Expand model catalog: 29 OCI models including Llama 4, Gemini 2.5, xAI Grok, Cohere Command A, and all Cohere embed variants - Add 37 live integration tests: sync/async completions for Meta/Google/xAI/Cohere, sync/async embeddings, tool use across all vendors, streaming, env var auth - Add 23 embed unit tests covering all transform and validation paths * fix(oci): remove dead OCI elif branch in utils.py, align async split_chunks with sync version * test(oci): add unit tests for split_chunks fix and no-duplicate-OCI-branch guard * fix(oci): address remaining bugs from issue #25082 — streaming signed body, Cohere stop sequences, hardcoded defaults - Bug 1: sync and async streaming paths now use signed_json_body when provided instead of re-serializing data with json.dumps() — the OCI RSA-SHA256 signature covers the exact request body bytes, so re-serializing produces an invalid sig - Bug 3: Cohere stop sequences now map to 'stopSequences' (was incorrectly 'stop') - Bug 4: removed hardcoded Cohere defaults (maxTokens=600, temperature=1, topK=0, topP=0.75, frequencyPenalty=0) that silently overrode user intent on every call - Added 6 unit tests covering all three fixes * fix(oci): comprehensive code quality pass — bugs, tests, schema accuracy - Fix Cohere tool call IDs (was always call_0; now UUID per call) - Fix TOOL_CALL finish reason mapping in both sync and streaming paths - Fix Cohere stop parameter mapping (stop → stopSequences) - Remove hardcoded Cohere defaults (maxTokens/topK/topP/frequencyPenalty) - Fix content[0] safety guard against empty content arrays - Fix streaming signed body used consistently (not re-serialized) - Raise OCIError (not bare Exception/ValueError) throughout - Centralize OCI_API_VERSION constant; import uuid at module level - Fix embed get_complete_url to strip trailing slashes from api_base - Fix OCIEmbedResponse schema: add inputTextTokenCounts (actual OCI field) - Fix embed usage computed from inputTextTokenCounts (sum of per-input counts) - Fix Cohere toolCallId included in tool result messages - Add OCIToolCall.id as Optional (absent in Google/xAI streaming chunks) - Update tests to reflect correct behavior (no hardcoded defaults, UUID ids, deferred credential validation, OCIError vs ValueError, real response schema) * test(oci): move integration tests to tests/llm_translation/ Addresses greptile P1: tests/test_litellm/ is for mock-only unit tests (make test-unit target). Real-network OCI tests now live in the correct location alongside other provider integration tests. * fix(oci): align types and transformation with official OCI SDK - Remove OCIVendors.GEMINI — apiFormat="GEMINI" is invalid; all non-Cohere models use apiFormat="GENERIC" - Add toolChoice, logitBias, logProbs to OCIChatRequestPayload so params present in the mapping are no longer silently dropped by Pydantic - Exclude n→numGenerations from Cohere param map (not a Cohere API field) - Fix CohereToolResult: change callId/result to call/outputs matching the OCI SDK's CohereToolResult structure - Fix CohereToolMessage: replace non-existent toolCallId with toolResults list; update adapt_messages_to_cohere_standard to build proper tool-result history entries by resolving tool call name+params from preceding assistant messages - Map generic-model stream finish reasons to OpenAI convention (COMPLETE→stop, MAX_TOKENS→length, TOOL_CALLS→tool_calls), consistent with the existing Cohere streaming path - Add optional id field to OCIEmbedResponse so valid API responses carrying an id are not rejected by the Pydantic model * fix(oci): use 'output' key in Cohere tool result outputs (matches reference impl) * fix(oci): port schema/type utilities from langchain-oracle reference impl - Add resolve_oci_schema_refs: inline $ref/$defs — OCI rejects JSON Schema refs - Add resolve_oci_schema_anyof: flatten Optional[T] anyOf (Pydantic v2 emits these) - Add sanitize_oci_schema: strip title, normalise null types, ensure array items - Add OCI_JSON_TO_PYTHON_TYPES: Cohere expects Python type names (str/int/float), not JSON Schema names (string/integer/number) - Add enrich_cohere_param_description: embed enum/format/range/pattern constraints into description since CohereParameterDefinition has no dedicated fields - Apply all of the above in adapt_tool_definitions_to_cohere_standard and adapt_tool_definition_to_oci_standard - Fix toolChoice conversion: map OpenAI string ('auto','none','required') to OCI dict form ({"type":"AUTO"} etc.) — the API rejects plain strings - Update unit test expectations to match correct Python type names and enriched descriptions * refactor(oci): split transformation.py into cohere.py and generic.py transformation.py was 1 243 lines doing too many jobs. Split along the same boundaries as the langchain-oracle reference (providers/cohere.py, providers/generic.py): chat/cohere.py — Cohere message/tool building, response + stream parsing chat/generic.py — Generic message/tool building, response + stream parsing transformation.py — thin OCIChatConfig orchestrator + OCIStreamWrapper Public symbols (OCIChatConfig, OCIStreamWrapper, adapt_messages_to_*, OCIRequestWrapper, version, …) remain importable from transformation.py for backward compatibility. OCIStreamWrapper gains delegating shims for _handle_cohere_stream_chunk and _handle_generic_stream_chunk so existing test call sites keep working unchanged. transformation.py: 1 243 → 620 lines * refactor(oci): principal-level code quality pass - Remove _extract_text_content duplication — single definition in cohere.py, imported where needed; instance method on OCIChatConfig eliminated - Move cryptography imports to module level with _CRYPTOGRAPHY_AVAILABLE flag and _require_cryptography() guard; no more re-import on every signing call - Move litellm version import to module level via litellm._version; remove inline import inside validate_oci_environment - sign_with_manual_credentials now returns Tuple[dict, bytes] matching sign_with_oci_signer — asymmetry eliminated, Optional[bytes] guards removed throughout stream wrappers (signed_json_body: bytes = b"") - Rename _openai_to_oci_cohere_param_map → openai_to_oci_cohere_param_map for consistency with openai_to_oci_generic_param_map - Remove double-key bug in map_openai_params where responseFormat was stored under both OCI and OpenAI key names simultaneously - Remove delegating shims (adapt_messages_to_cohere_standard, adapt_tool_definitions_to_cohere_standard, _handle_generic_stream_chunk) from OCIChatConfig/OCIStreamWrapper; tests now import directly from cohere.py and generic.py where symbols live - Trim __all__ to 7 genuine public symbols; remove the 13-symbol list that existed only to support test imports - Collapse per-model integration test classes into pytest.mark.parametrize; CHAT_MODELS list is the single source of truth for model-specific config - Black + Ruff clean across all OCI files * fix(oci): address PR review findings - types/llms/oci.py: add "TOOL_CALL" to CohereChatResponse.finishReason Literal so Pydantic does not raise ValidationError on non-streaming Cohere tool-use calls (Greptile P1) - test_oci_cohere_tool_calls.py: add test covering TOOL_CALL finish reason - model_prices_and_context_window.json: remove 6 duplicate oci/cohere.embed-* keys that were silently overridden by the more complete entries already present in the file (Greptile P1) - common_utils.py: move OCI_API_VERSION here from chat/transformation.py so embed/transformation.py does not need to import chat/transformation; change Protocol stub body from ... to pass (CodeQL "statement no effect"); add comment to sha256_base64 clarifying it implements OCI HTTP signing spec, not password hashing (CodeQL false positive) - chat/transformation.py: import CustomStreamWrapper from litellm_core_utils.streaming_handler instead of litellm.utils to reduce import cycle depth (CodeQL cyclic import) - chat/cohere.py, chat/generic.py: import Usage and ChatCompletionMessageToolCall from litellm.types.utils instead of litellm.utils for the same reason - embed/transformation.py: import OCI_API_VERSION from common_utils instead of chat/transformation (removes the embed→chat import edge) * test(oci): add unit tests to improve patch coverage - test_oci_common_utils.py (new): covers sha256_base64, build_signature_string, OCIRequestWrapper.path_url, resolve_oci_credentials, get_oci_base_url, validate_oci_environment, sign_with_oci_signer error paths, sign_oci_request routing, load_private_key_from_file error paths, resolve_oci_schema_refs (including circular ref and external $ref), resolve_oci_schema_anyof, sanitize_oci_schema (all branches), enrich_cohere_param_description - test_oci_generic_chat.py (new): covers content-message error paths (non-dict item, unsupported type, non-string text, invalid image_url), tool-call validation error paths, adapt_messages_to_generic_oci_standard error paths, handle_generic_response (None message, text content, tool calls), handle_generic_stream_chunk (finish reasons, streaming tool calls), OCIStreamWrapper non-string chunk error - test_oci_chat_transformation.py: add error paths for validate_environment (empty messages), transform_request (missing compartment_id, Cohere without user messages), transform_response (error key), map_openai_params (unsupported param with and without drop_params), tool_choice string mapping - test_oci_cohere_tool_calls.py: add edge cases for stream chunk finish reasons (TOOL_CALL, MAX_TOKENS, unknown), _extract_text_content with non-dict list items and non-string input, adapt_messages_to_cohere_standard with malformed JSON tool arguments * fix(oci): rename supports_streaming to supports_native_streaming in model prices The JSON schema for model_prices_and_context_window.json uses `supports_native_streaming` (not `supports_streaming`) and has `additionalProperties: false`. Rename the field across all OCI entries to pass the schema validation test. * test(oci): add 67 tests targeting uncovered happy paths for coverage Boost patch coverage on the four lowest-coverage OCI files: - common_utils.py: sign_with_manual_credentials (oci_key / oci_key_file paths), sign_oci_request routing, _require_cryptography - generic.py: adapt_messages_to_generic_oci_standard (all roles), adapt_tool_definition_to_oci_standard, adapt_tools_to_openai_standard, handle_generic_stream_chunk text/finish-reason paths - cohere.py: _extract_text_content, adapt_messages_to_cohere_standard (all roles including tool results), handle_cohere_response / handle_cohere_stream_chunk all finish-reason branches - transformation.py: get_vendor_from_model, OCIChatConfig._get_optional_params (toolChoice string→dict, responseFormat, tools for both vendors), transform_request for GENERIC model, get_sync/async_custom_stream_wrapper with mocked HTTP, OCIStreamWrapper.chunk_creator happy paths * fix(oci): suppress CodeQL false positive on sha256_base64 (OCI HTTP signing, not password hashing) * fix(oci): remove 6 duplicate model price entries and reconcile conflicting values Six OCI chat model keys appeared twice in model_prices_and_context_window.json with conflicting pricing/context data (JSON parsers silently discard the first). Remove the first-occurrence entries and update the surviving entries: - meta.llama-4-maverick / llama-4-scout: keep updated entries (free preview pricing, larger context windows, vision support) - meta.llama-3.1-70b: keep original pricing, restore supports_native_streaming - google.gemini-2.5-{flash,pro,flash-lite}: keep OCI pricing page values, restore supports_native_streaming * fix(oci): route GPT-5 family to maxCompletionTokens GPT-5 / GPT-5-mini / GPT-5-nano / GPT-5.5 on OCI reject "maxTokens" with HTTP 400: Invalid 'maxTokens': Unsupported parameter: 'maxTokens' is not supported with this model. Use 'maxCompletionTokens' instead. (Same convention as OpenAI's reasoning-API contract.) Add a model-aware rename in OCIChatConfig._get_optional_params so the request payload uses maxCompletionTokens when the model id starts with openai.gpt-5. Regular Llama / Cohere / Gemini / GPT-4.x continue to use maxTokens unchanged. Also widen OCIChatRequestPayload to carry the new optional field so it survives Pydantic serialization. Verified live against OCI us-chicago-1: - openai.gpt-5, gpt-5-mini, gpt-5-nano, gpt-5.5 all return 200 - Full feature sweep on gpt-5.5 (basic, system, multi-turn, streaming, tools, usage) all green - meta.llama-3.3-70b-instruct still uses maxTokens (no regression) 4 new unit tests cover the helper, the routing in both pre- and post-translation states, and Pydantic serialization. * ci(oci): fix CI failures — black formatting + recursive_detector ignore - Run black on litellm/llms/oci/common_utils.py + 3 OCI test files that drifted out of black-compliance during the rebase. - Add the three bounded recursive functions in oci/common_utils.py (`_resolve`, `resolve_oci_schema_anyof`, `sanitize_oci_schema`) to the recursive_detector IGNORE_FUNCTIONS list. All three are bounded: `_resolve` uses a `resolving_stack` cycle guard; the other two are bounded by JSON-schema tree depth (no cycles in well-formed input), matching the pattern of the existing OCI/Vertex schema walkers already on the list. * fix(oci): silence MyPy errors in cohere.py — typed-dict access Two errors flagged by `lint` CI: llms/oci/chat/cohere.py:73: "object" has no attribute "__iter__" llms/oci/chat/cohere.py:119: No overload variant of "get" of "dict" matches argument types "object", "CohereToolCall" Both stem from `msg.get("tool_calls")` / `msg.get("tool_call_id")` returning `object` per the AllMessageValues TypedDict union. Bind to `Any` locally for the iteration and coerce the lookup key with `str()`, removing the now-unused `# type: ignore` on those lines. No behaviour change — pure type-narrowing for the type checker. * fix(oci): silence CodeQL py/weak-sensitive-data-hashing on sha256_base64 CodeQL's taint analysis traces request bodies back to environment-loaded secrets and flags `hashlib.sha256(body).digest()` as `py/weak-sensitive-data-hashing` — even though SHA-256 is the algorithm mandated by the OCI HTTP request signing spec for the `x-content-sha256` header (not a password/secret hash). The previous suppression used legacy `# lgtm[...]` syntax which the modern CodeQL action ignores. Switch to Python's standard `hashlib.sha256(..., usedforsecurity=False)` (Python 3.9+) which CodeQL honours as a non-security declaration. Behaviour unchanged. * feat(oci): add reasoning_effort passthrough — only true missing primitive OCI's GenericChatRequest exposes a reasoningEffort field (NONE/MINIMAL/LOW/MEDIUM/HIGH) that's the single biggest cost knob for reasoning-capable models on the service: - GPT-5 family - Gemini 2.5 - Grok reasoning variants (3-mini, 4-fast, 4.20) - Cohere Command-A-Reasoning Setting reasoning_effort=LOW typically cuts reasoning-token spend 5-10× vs the default. Without exposing this, litellm users had no way to tune cost-vs-quality on these models. The other GenericChatRequest fields (verbosity, parallel_tool_calls, logit_bias, n, metadata, web_search_options, prediction) are not exposed because they are not missing primitives — they either duplicate prompt-engineering, framework-level controls, or are too niche to justify the maintenance surface. We only ship what users genuinely can't accomplish another way. Excluded from the Cohere v1 param map: CohereChatRequest has no reasoningEffort field, and Cohere reasoning models (cohere.command-a-reasoning) use COHEREV2 which is a separate request type not covered by this PR. Verified live: GPT-5.5 + reasoning_effort="HIGH" sends {"reasoningEffort": "HIGH"} on the wire and OCI accepts the request. * feat(oci): reasoning_effort + reasoning_tokens for OCI GenAI Three small additions for OCI reasoning models, requested by users testing the PR in production fork builds: 1. **reasoning_effort param mapping (GENERIC vendors).** OCI expects uppercase levels ("LOW"/"MEDIUM"/"HIGH"/"NONE") on `reasoningEffort`, but OpenAI-compatible clients send lowercase. Mapped + uppercased in `_get_optional_params`. Marked unsupported on Cohere V1/V2 since OCI Cohere has no reasoning models (avoids Pydantic validation failure on CohereChatRequest). 2. **"disable" → "NONE" mapping.** OpenAI uses "disable" to turn off reasoning; OCI uses "NONE". Without this, callers get a 400. 3. **reasoning_tokens propagated to Usage.** OCI returns `completionTokensDetails.reasoningTokens` but it wasn't being passed to LiteLLM's Usage object. Now flows through to `Usage.completion_tokens_details.reasoning_tokens` so callers can track reasoning token consumption for cost/observability. Tests: 7 new unit tests in TestOCIReasoningEffort covering upper/lower case, "disable"→"NONE", Cohere drop/raise paths, and reasoning_tokens extraction (with and without completionTokensDetails). 5 new live integration tests against xai.grok-3-mini in us-chicago-1 verifying the full request/response loop end-to-end. Existing test_transform_response_simple_text assertion that completion_tokens_details was None has been updated to assert reasoning_tokens flows through. Verified live on xai.grok-3-mini: reasoning_effort=low → OCI accepts "LOW", returns reasoningTokens=316 in usage. reasoning_effort=disable → OCI accepts "NONE". Full suite: 370/370 unit + 51/51 integration. * fix(codeql): re-scope py/weak-sensitive-data-hashing exclusion to OCI signing file CodeQL's taint analysis re-fires the `py/weak-sensitive-data-hashing` alert at `litellm/llms/oci/common_utils.py:103` whenever upstream code paths into the OCI signing module change (touching `transformation.py` opens new flow paths that CodeQL re-evaluates from scratch). The `hashlib.sha256(..., usedforsecurity=False)` declaration silences the direct-call form of the query but not the taint-flow form. SHA-256 here is mandated by the OCI HTTP signing specification for the x-content-sha256 content-integrity header — not for password storage: https://docs.oracle.com/en-us/iaas/Content/API/Concepts/signingrequests.htm CodeQL has no per-query path filter and GitHub Code Scanning ignores inline lgtm/codeql comments, so path-ignoring this single ~560-line signing utility file is the narrowest available suppression. All other files retain full coverage of py/weak-sensitive-data-hashing — including litellm/proxy/utils.py where the rule legitimately applies. This restores the NEUTRAL CodeQL state the PR had on prior commits (see `2111c98af7` for the same approach on the previous branch evolution that the cherry-pick was rebased onto a different baseline). * fix(oci): drop duplicate text on Cohere streaming terminal chunk OCI Cohere's terminal SSE event re-sends the full assembled response in `text` alongside a populated `chatHistory`. Emitting that text as another delta concatenates the entire response onto the already-streamed output (e.g. "How can I help?How can I help?"). Use `chatHistory is not None` as the discriminator for the consolidated terminal event — `finishReason` is a weaker signal that could in principle appear on a non-consolidated chunk. The two coincide today; this preserves correctness if OCI ever ships finishReason on an incremental chunk. Adds a live-OCI integration regression test that compares streamed vs non-streamed length and asserts the response prefix appears only once. Verified to fail under the previous code with the exact reported reproduction: 'Hello! How can I help you today?Hello! How can I help you today?'. Reported by @gotsysdba on PR #25177. * fix(oci): buffer SSE stream across HTTP read boundaries The old split_chunks helper split each individual HTTP read on "\n\n", which assumed SSE event boundaries always aligned with read boundaries. In practice the OCI streaming endpoint delivers events that may: - straddle two reads (chunk_creator gets a truncated JSON and crashes) - arrive separated by a single "\n" instead of "\n\n" - share a read with multiple complete events Replace the inline split with module-level helpers _iter_sse_events (sync) / _aiter_sse_events (async) that maintain a buffer across reads, split on any newline, and yield only complete "data:" lines. Add 25 regression tests covering event-split-across-reads, tiny-chunk reads, single-newline separators, keepalive/comment lines, trailing partial events flushed at EOF, "\r\n" line endings, and an end-to-end smoke test that feeds an awkwardly-chopped payload through the splitter into OCIStreamWrapper.chunk_creator. Reported by John Lathouwers. * test(oci): repoint TestOCIKeyNormalization to sign_with_manual_credentials The signing helper moved from OCIChatConfig._sign_with_manual_credentials to a module-level sign_with_manual_credentials in common_utils.py. Four tests in TestOCIKeyNormalization still called the old method: - 2 failed outright with AttributeError - 2 passed by accident because they used pytest.raises(Exception), which happily caught the AttributeError instead of exercising the intended OCIError path Repoint all four to the new module-level function so they exercise the actual oci_key type-validation branch. * fix(oci): validate oci_region before URL interpolation to prevent SSRF Anchor oci_region to ^[a-z][a-z0-9-]{0,30}[a-z0-9]$ inside get_oci_base_url so user-supplied regions that would redirect the signed request to an attacker-controlled host (e.g. 'evil.com/#') fail with HTTP 400 before the URL or signature is built. Empty string still falls back to the us-ashburn-1 default, so existing callers are unaffected. * test(audio): skip when gpt-4o-audio-preview is unavailable upstream OpenAI retired `gpt-4o-audio-preview` (404 model_not_found in CI as of 2026-05-19), and the existing try/except in these tests only re-raised on 'openai-internal' errors. Other exceptions were silently swallowed, so the next line ran with an unbound `response`/`completion` and failed with an unrelated UnboundLocalError that masked the real cause. Extend the skip condition to also cover model_not_found / 'does not exist' so the suite reports the upstream outage cleanly, matching the pattern used in ce87c41 for the realtime and nvidia_nim rerank tests. Re-raise unknown exceptions instead of falling through. * fix(oci/router): catalog-driven maxCompletionTokens; generic blocked-deployment message - Drive OCI maxCompletionTokens via supports_reasoning from the model catalog instead of a hardcoded openai.gpt-5 prefix. Add OCI GPT-5 family entries (gpt-5, gpt-5-mini, gpt-5-nano) with supports_reasoning: true. Gate the override to non-Cohere vendor so Cohere reasoning models keep maxTokens (Cohere endpoint does not accept maxCompletionTokens). - Replace proxy-specific 'Contact your proxy admin' phrasing in the four Router blocked-deployment ServiceUnavailableError messages with neutral SDK-appropriate text. * fix(oci/cohere): guard handle_cohere_response against missing usage * fix(oci): address bug review findings in chat transformation - Cohere param map: keep tool_choice/n as False (not omitted) so unsupported params are dropped or rejected rather than silently passed through. - get_complete_url: when an explicit api_base/litellm.api_base is provided, use it as-is instead of unconditionally appending /20231130/actions/chat (mirrors the embed config behavior). - Cohere stream: require both chatHistory and finishReason to be present to identify a terminal consolidation chunk, avoiding silent text suppression if chatHistory ever appears on a non-terminal chunk. - Generic usage: use 'is not None' for reasoningTokens so a legitimate value of 0 is preserved instead of being treated as absent. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci/cohere): emit tool calls in streaming and null content when text empty handle_cohere_response now sets message.content to None when the Cohere response text is empty, matching the OpenAI convention for tool-call-only responses. handle_cohere_stream_chunk now extracts toolCalls — both directly from the chunk and from the terminal chunk's chatHistory CHATBOT message — and emits them in the delta. Previously, CohereStreamChunk lacked a toolCalls field, so any tool calls in the stream were silently dropped. * fix(oci): preserve tool results, embed URL path, and generic finish reason - Use SerializeAsAny on CohereChatRequest.chatHistory so subclass-specific fields like CohereToolMessage.toolResults are not dropped during Pydantic v2 serialization. - Make OCIEmbedConfig.get_complete_url append the /20231130/actions/embedText action path consistently with chat, so setting litellm.api_base to the region inference base URL no longer posts to the bare hostname. - Map OCI finishReason (COMPLETE / MAX_TOKENS / TOOL_CALLS) to OpenAI finish_reason values in handle_generic_response, mirroring the streaming handler and the Cohere non-streaming handler. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci/generic): silence mypy assignment error on dynamic finish_reason * fix(oci/embed): always set usage on embedding response Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci/chat): append /20231130/actions/chat to explicit api_base Restore the embed-style behavior so OCIChatConfig.get_complete_url always appends the OCI GenAI chat path. Routing through get_oci_base_url ensures the optional explicit api_base has its trailing slash stripped before the suffix is joined, matching the embed config and the test_respects_explicit_api_base expectation. * fix(oci/cohere): mark logprobs/logit_bias unsupported and normalize unknown stream finish reasons Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci/cohere): preserve trailing tool result in chatHistory When the last message in the OpenAI-format input is a tool result (the standard agentic continuation pattern), the prior messages[:-1] slice silently dropped that tool result from chatHistory and the model never saw it. Excluding the last user message by index instead keeps tool results that trail the last user turn intact. * fix(main): remove dead OCI embedding elif block The earlier elif at line 5119 already routes OCI embeddings through the base HTTP handler with the headers None-guard, so the later identical block was unreachable dead code. * test(oci): move integration tests out of llm_translation mock-only folder Greptile flags tests/llm_translation/ as mock-only via a project-specific rule; relocate the live-network OCI integration suite to tests/integration/ and adjust the in-file sys.path / run instructions accordingly. * fix(oci/cohere): suppress tool calls on stream terminal consolidation chunk The terminal SSE event re-sends the full assembled response in both `text` and `chatHistory`. The existing logic already suppresses `text` to avoid double-emit, but tool calls extracted from the terminal chunk (via `typed_chunk.toolCalls` or the `chatHistory` CHATBOT fallback) would still be re-emitted with fresh uuid4 IDs. If OCI Cohere ever streams tool calls progressively in intermediate chunks (now possible since CohereStreamChunk has a toolCalls field), this would cause downstream agentic frameworks to execute each tool call twice. Suppress tool calls on the terminal consolidation chunk for the same reason `text` is suppressed. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci,httpx): normalize finish_reason, preserve response_format, fix sync embed JSON content-type - cohere.py / generic.py: normalize unknown OCI finishReason values (ERROR, ERROR_TOXIC, CONTENT_FILTERED, USER_CANCEL, ...) to 'stop' in non-streaming and streaming generic handlers, matching the streaming Cohere handler so downstream consumers switching on finish_reason aren't broken by raw OCI values. - transformation.py: restore the dual-key alias so optional_params still carries the original 'response_format' key alongside the OCI-mapped 'responseFormat'. Downstream litellm framework code (json_mode detection, logging) inspects 'response_format' after map_openai_params runs. - llm_http_handler.py: make the sync embedding path mirror the async path — when sign_request returns no signed_body, send via json=data (which sets Content-Type: application/json) instead of data=json.dumps(data) which doesn't. Removes a sync/async behavioural asymmetry for non-OCI providers that adopt the sign_request pattern. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci): clean up OCIChatConfig init, normalize generic stream finish reasons, correct embed sign_request return type - Replace fragile setattr(self.__class__, ...) pattern in OCIChatConfig.__init__ with a @property for has_custom_stream_wrapper, matching the pattern used by other providers. - Normalize unknown OCI finish reasons (e.g. ERROR, ERROR_TOXIC, USER_CANCEL) to 'stop' in handle_generic_stream_chunk, matching the existing Cohere stream handler behaviour. - Tighten OCIEmbedConfig.sign_request return type from Tuple[dict, Optional[bytes]] to Tuple[dict, bytes] — sign_oci_request never returns None for the body, and this matches OCIChatConfig.sign_request. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci): strip trailing action path in get_oci_base_url to avoid URL doubling A fully-formed OCI endpoint URL (e.g. https://inference.generativeai.us-chicago-1.oci.oraclecloud.com/20231130/actions/chat) passed via api_base previously had the action path appended a second time by get_complete_url in both chat and embed configs, yielding a 404. get_oci_base_url now strips a trailing /20231130/actions/<name> so callers can always append the action path safely. * fix(httpx): preserve sync embed data= kwarg to avoid breaking mock-based tests The earlier sync_httpx_client.post() call passed data=json.dumps(data), which downstream embedding tests assert on (e.g. tests for hosted_vllm, jina_ai, watsonx). Switching to json=data changed the kwarg name and broke those tests. The OCI signed_body path keeps using data=signed_body and is unaffected. * fix(oci): stable tool-call ids across stream chunks; lenient Cohere finishReason - Replace random uuid4 per chunk with a deterministic content-derived digest for synthetic tool-call ids in both Cohere and Generic OCI handlers. Previously, when OCI omitted 'id' (always for Cohere, often for Generic streaming deltas), every chunk for the same logical tool call received a new uuid, causing downstream stream-mergers (which key off id) to treat each fragment as a distinct call. - Relax CohereChatResponse.finishReason from a strict Literal[...] to Optional[str], matching CohereStreamChunk.finishReason. The handle_cohere_response 'elif oci_finish_reason is not None' fallback was previously unreachable because Pydantic raised ValidationError on any unknown value before the fallback executed. Now non-streaming responses degrade unknown reasons to 'stop' just like the streaming path. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci/embed): validate OCI credentials in validate_environment Mirror OCIChatConfig.validate_environment so embedding requests fail fast with a clear error when oci_user/oci_fingerprint/oci_tenancy/ oci_compartment_id or an oci_key/oci_key_file is missing, instead of deferring the failure until sign_request. Co-authored-by: Yassin Kortam <yassin@berri.ai> * test(oci/embed): expect OCIError from validate_environment when credentials are missing OCIEmbedConfig.validate_environment now raises eagerly (mirroring OCIChatConfig) when oci_user/oci_fingerprint/oci_tenancy/oci_compartment_id or oci_key/oci_key_file is missing. Update the test to match. * fix(oci): polish stream chunk handling and signed body default - cohere stream terminal consolidation now emits content=None instead of "" - drop redundant index truthiness check (None is already replaced with 0) - accept both "TOOL_CALL" and "TOOL_CALLS" finish reasons in cohere - signed_json_body defaults to None and uses explicit None check, so an explicitly empty bytes body wouldn't be silently re-serialized Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci/chat): catch pydantic ValidationError when parsing OCI responses Pydantic v2 raises ValidationError (not TypeError) when field validation fails, so malformed OCI completion responses or stream chunks would propagate unhandled out of handle_generic_response, handle_generic_stream_chunk, and handle_cohere_stream_chunk. Widen the except clauses to also catch ValidationError so callers get a clean OCIError. * fix(oci/catalog): real prices for Llama 4, drop zero-cost OCI OpenAI entries Zero-cost catalog entries (input_cost_per_token=0, output_cost_per_token=0) make proxy spend tracking silently report $0 for these paid OCI models, so any caller can drive them without decrementing a budget. For Llama 4 Maverick and Scout, OCI charges the same character-based rate as Llama 3.3 70B ($0.0018 per 10,000 characters), so use the same per-token price as the existing oci/meta.llama-3.3-70b-instruct entry (7.2e-07 in/out). For oci/openai.gpt-5, gpt-5-mini, gpt-5-nano, gpt-oss-120b, and gpt-oss-20b, no public per-token pricing is available; drop the entries so operators must register them with explicit custom pricing. The existing GPT-5 reasoning test fixture already injects synthetic entries when the catalog omits them, so the chat transformation's supports_reasoning lookup keeps working in tests. * fix(oci/chat): wrap CohereChatResult construction in try/except Match the handle_generic_response pattern: surface OCIError with the upstream status code instead of letting a raw pydantic.ValidationError propagate when the Cohere response payload is malformed. * fix(oci): harden Cohere stream/finish-reason and dedupe maxTokens param mapping - Cohere stream: track per-stream tool-call emission and only suppress the terminal consolidation chunk's tool calls once they've been seen earlier. Prevents silent drop if tool calls are delivered exclusively on the terminal chunk. - Cohere stream: emit content=None (not "") on non-terminal text-free chunks (e.g. tool-call-only / keep-alive) so downstream consumers that distinguish missing vs explicitly-empty deltas behave correctly. - Generic handlers: accept singular TOOL_CALL finish reason in addition to TOOL_CALLS, matching the Cohere handlers. - _get_optional_params: when both max_tokens and max_completion_tokens are provided, explicitly prefer max_completion_tokens instead of relying on dict iteration order. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci): emit content=None instead of empty string for text-free generic stream chunks Co-authored-by: Yassin Kortam <yassin@berri.ai> * test(oci): expect content=None for text-free generic stream chunks handle_generic_stream_chunk now emits content=None instead of empty string when a chunk carries no text parts. Update the corresponding no-message test to match. * codeql: narrow OCI sha256 suppression to query-filter, not whole file paths-ignore was suppressing every CodeQL query on litellm/llms/oci/common_utils.py, hiding all future findings in a security-critical file (private key loading, credential resolution, URL construction, RSA signing). Move the suppression for py/weak-sensitive-data-hashing into query-filters so common_utils.py remains fully analyzed by every other query. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci): use locale-independent RFC 7231 date for manual signing email.utils.formatdate(usegmt=True) emits canonical English weekday/ month abbreviations regardless of system locale, so signature verification doesn't break on non-en_US deployments. * fix(oci): strip 'oci/' prefix in get_vendor_from_model Previously, get_vendor_from_model split on '.' without stripping the optional 'oci/' provider prefix, so 'oci/cohere.command-a-03-2025' was routed through the GENERIC pipeline instead of COHERE. Co-authored-by: Yassin Kortam <yassin@berri.ai> * codeql: scope OCI sha256 suppression to common_utils.py via filter-sarif Replace the global query-filters exclude for py/weak-sensitive-data-hashing with a SARIF post-filter that only drops the alert when it originates from litellm/llms/oci/common_utils.py, keeping the rule active on every other SHA-256 callsite in the repository. * Fix OCI chat bugs: tool_calls None key, dead max_tokens dedup, single-event stream text suppression - handle_cohere_response: omit tool_calls key from message dict when None, matching the generic handler's behaviour and avoiding tripping consumers that key off 'tool_calls' in message. - _get_optional_params: remove dead prefer_max_completion branch. By the time this helper runs, map_openai_params has already collapsed max_tokens/max_completion_tokens onto the OCI alias, so the OpenAI-key membership check is unreachable. - handle_cohere_stream_chunk: add prior_text_emitted parameter mirroring prior_tool_calls_emitted. The terminal consolidation chunk's text is only suppressed when prior deltas already emitted text — otherwise (degenerate single-event stream) the text passes through so the response content isn't silently lost. OCIStreamWrapper now tracks emitted text alongside emitted tool calls. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci): preserve all text parts in generic response and emit SYSTEM role for Cohere - handle_generic_response: iterate all content parts and concatenate text (matches the streaming handler) so non-leading text parts are not lost and a leading non-text part does not suppress trailing text. - adapt_messages_to_cohere_standard: emit CohereSystemMessage for system messages so direct callers do not silently drop them. The Cohere request builder filters system messages before calling this helper to avoid duplicating preambleOverride content into chatHistory. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci): normalise dict-format tool_choice to OCI flat uppercase shape The OCI Generative AI API only accepts toolChoice values of the form {"type": "AUTO"|"NONE"|"REQUIRED"} or {"type": "FUNCTION", "name": "<fn>"}. The previous conversion only handled string tool_choice values, so OpenAI's standard dict shape {"type": "function", "function": {"name": "<fn>"}} passed through unchanged and was rejected by OCI with a 400. Normalise the dict shape by uppercasing the discriminator and hoisting the function name to the top level. Also accept dict variants of the non-function selectors (e.g. {"type": "auto"}). * test(oci): exercise system-message filtering at transform_request boundary adapt_messages_to_cohere_standard now emits SYSTEM-role entries by design so direct callers don't silently drop system content. The Cohere request builder filters system messages before calling the helper and routes them into preambleOverride, so the user-visible 'no SYSTEM in chatHistory' guarantee holds at the transform_request boundary, where the test should live. * fix(oci/chat): extract tool_choice/response_format helpers to satisfy PLR0915 _get_optional_params exceeded ruff's 50-statement cap. The toolChoice and responseFormat normalisation blocks are self-contained mutations, so move them to module-level helpers. * fix(oci): normalize None finishReason in generic non-streaming handler; drop dead Cohere system-role branch Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci/generic): silence mypy assignment error on cleared finish_reason * fix(docker): install libatomic in builder for prisma nodeenv binary The prebuilt node binary that prisma-python's nodeenv downloads links against libatomic.so.1, which Wolfi does not pull in via gcc/nodejs. Without this, fresh Docker builds (no GHA cache hit) fail at `prisma generate` with: node: error while loading shared libraries: libatomic.so.1 * fix(oci): raise on invalid tool_choice instead of silently passing OpenAI shape _normalize_tool_choice previously left an OpenAI-format dict in selected_params['toolChoice'] when the type was unrecognized or when 'FUNCTION' was given with a missing/empty name. OCI would then reject the request with a non-obvious error. Raise ValueError with a clear message in these cases. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci): raise OCIError instead of ValueError in _normalize_tool_choice Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci/generic): declare non-security intent on sha256 for synthetic tool-call id * fix(oci): simplify _get_optional_params and reject invalid tool_choice types - Collapse the two-loop _get_optional_params into a single pass with clear precedence (OpenAI key wins over OCI alias; first OpenAI key reaching a given OCI target wins). Removes the redundant maxTokens special-case in the second loop and makes the map_openai_params / transform_request handoff easier to reason about. - Raise OCIError when _normalize_tool_choice sees an unexpected type (list, bool, int, ...) instead of silently letting it through to the OCI API where it would produce an opaque server-side error. Co-authored-by: Yassin Kortam <yassin@berri.ai> * Remove no-op data['stream'] deletion in OCI stream wrappers Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci): always send Cohere isStream field explicitly Match OCIChatRequestPayload by defaulting CohereChatRequest.isStream to False instead of None so model_dump(exclude_none=True) does not silently omit the field on non-streaming requests. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci): revert Cohere isStream to Optional[bool]=None to preserve omission semantics Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci/generic): raise OCIError on empty choices instead of IndexError Pydantic accepts an empty choices list when validating OCICompletionResponse, so accessing chatResponse.choices[0] could raise an unhandled IndexError. Surface it as OCIError so the response error path is consistent with the existing (TypeError, ValidationError) guard. * fix(oci/cohere): map top_k -> topK so Cohere topK param is settable The Cohere param map (derived from the GENERIC map) had no entry for topK. Since the simplified _get_optional_params only iterates over param_map entries, callers had no way to pass topK to CohereChatRequest (neither via an OpenAI-style key nor via the OCI alias). Add 'top_k': 'topK' to the Cohere map only — OCIChatRequestPayload (GENERIC) has no topK field. _get_optional_params accepts both the OpenAI key (top_k) and the OCI alias (topK) in optional_params, so this covers both calling conventions. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci): tighten cohere stream dedup flags and forward stream args in embed signing Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci/chat): reorder dict guard and wrap stream chunk json.loads - Move isinstance(response_json, dict) check before .get("error") so the guard runs before the attribute access it is supposed to protect. - Wrap json.loads in OCIStreamWrapper.chunk_creator with try/except so malformed SSE payloads surface as OCIError instead of a raw JSONDecodeError propagating out of the stream loop. * fix(oci/cohere stream): only flag text emitted on non-empty content An intermediate Cohere SSE chunk carrying text="" was flipping _cohere_text_emitted via the "is not None" check, which then caused the terminal consolidation chunk to drop its real text as a duplicate. Use a truthy check so only actual content marks the stream as having emitted text. * test(oci): end-to-end proxy integration test against real OCI GenAI Spins up the litellm proxy via the console-script entrypoint with a minimal OCI-only config and drives real OpenAI-shaped HTTP requests through it against OCI GenAI. Covers non-streaming chat, streaming chat, embeddings, and /v1/models for Cohere, Llama, Gemini, and Grok. Skips automatically when ~/.oci/config is absent or when the active profile uses session-token auth (the OCI provider currently only consumes OCI_* env vars; session tokens would need an in-process signer). API-key profiles work out of the box. * test(oci): move proxy integration test to tests/integration/ tests/llm_translation/ is mock-only; the OCI proxy integration test spawns a real proxy subprocess and makes live HTTP calls, so move it (and the companion config) to tests/integration/ alongside the existing test_oci_integration.py. * fix(oci): dedupe finish-reason mapping and batch Cohere tool results - Extract _normalize_oci_finish_reason helper so the four chat handlers (Cohere/GENERIC, sync/stream) share one OCI->OpenAI mapping instead of four near-identical if/elif chains. - Merge consecutive OpenAI tool-role messages into a single CohereToolMessage with multiple toolResults entries, matching the OCI Cohere API's expectation for parallel tool calls in one assistant turn. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(oci): drop dead Cohere toolChoice field and emit GENERIC tool-call dicts inline - Remove the unreachable toolChoice field from CohereChatRequest. The Cohere param map explicitly marks tool_choice as unsupported, so the field can never be populated through the normal optional_params flow and only confused the public model surface. - Build GENERIC stream tool-call dicts inline (id/type/function shape) instead of round-tripping through ChatCompletionMessageToolCall and model_dump(). Matches handle_cohere_stream_chunk so downstream stream-mergers see the same minimal payload regardless of which vendor produced the chunk. Co-authored-by: Yassin Kortam <yassin@berri.ai> * fix(docker): drop redundant libatomic from non_root builder litellm_internal_staging already fixes the prisma `nodeenv` build failure at the root cause by restoring `npm` to the builder (#28519): with npm on PATH, prisma-python uses the system Node and never downloads the nodeenv binary that links against libatomic.so.1. After merging internal_staging the libatomic line is dead weight, so remove it. https://claude.ai/code/session_01SwKzxRxgUhLFyyEf4UV812 * fix(oci/catalog): add openai.gpt-5{,-mini,-nano} entries with supports_reasoning Without these catalog entries, supports_reasoning(model='openai.gpt-5*', custom_llm_provider='oci') returned False, so _model_uses_max_completion_tokens fell back to the default and OCI rejected the request with HTTP 400 ('Use maxCompletionTokens instead.'). Add the three entries so the catalog-driven maxCompletionTokens routing works against a stock LiteLLM install. Also reword the test fixture docstring — the bundled backup now actually ships these entries, so the fixture is only a fallback for environments that loaded their cost map from a stale remote source. --------- Co-authored-by: Tai An <antai12232931@outlook.com> Co-authored-by: Vincent <yimao1231@gmail.com> Co-authored-by: Kris Xia <xiajiayi0506@gmail.com> Co-authored-by: d 🔹 <liusway405@gmail.com> Co-authored-by: Fabrizio Cafolla <developer@fabriziocafolla.com> Co-authored-by: Filippo Menghi <113345637+Cyberfilo@users.noreply.github.com> Co-authored-by: Tom Denham <tom@tomdee.co.uk> Co-authored-by: escon1004 <70471150+escon1004@users.noreply.github.com> Co-authored-by: Divyansh Singhal <97736786+Divyansh8321@users.noreply.github.com> Co-authored-by: robin-fiddler <robin@fiddler.ai> Co-authored-by: Michael-RZ-Berri <michael@berri.ai> Co-authored-by: Michael Riad Zaky <michaelr@Mac.localdomain> Co-authored-by: ryan-crabbe-berri <ryan@berri.ai> Co-authored-by: Krrish Dholakia <krrish+github@berri.ai> Co-authored-by: Sameer Kankute <sameer@berri.ai> Co-authored-by: Cursor <cursoragent@cursor.com> Co-authored-by: Yassin Kortam <yassin@berri.ai> Co-authored-by: Federico Kamelhar <federico.kamelhar@oracle.com> Co-authored-by: Claude <noreply@anthropic.com>	2026-05-23 12:15:41 -07:00
user	4d92bc8b86	fix(vector-stores): re-raise HTTPException from get_vector_store_info; allowlist recursion ... Two issues from the previous push's review: 1. **Greptile P1**: ``get_vector_store_info`` had the same catch-all ``except Exception`` pattern as ``update_vector_store``, so the HTTPException(403/404) raised by both the in-memory access check and the new ``_fetch_and_authorize_vector_store`` helper was rewritten as 500. Mirror the ``except HTTPException: raise`` guard from ``update_vector_store``. 2. **code-quality CI** (``tests/code_coverage_tests/recursive_detector.py``) flagged ``_redact_sensitive_litellm_params`` as an unallowlisted recursive function. Match the convention of other allowlisted helpers ("max depth set"): bound recursion at depth 10 (well above any plausible nesting level for real ``litellm_params`` payloads), return the redaction sentinel on overflow, and add the function name to ``IGNORE_FUNCTIONS``. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>	2026-04-29 18:56:55 +00:00
Yuneng Jiang	e7f4e77af0	Bound _get_masked_values recursion depth ... Add _depth/_max_depth guards (default 20) so the nested dict masking cannot run away, and allowlist the function in the recursive_detector CI check alongside the other bounded recursive helpers.	2026-04-24 13:00:33 -07:00
yuneng-jiang	a93c069dd5	[Fix] Add max_depth guard to BFL _read_image_bytes recursive function ... Use the standard depth/max_depth pattern with DEFAULT_MAX_RECURSE_DEPTH to guard the recursive list-unwrapping in _read_image_bytes, matching the existing pattern used by _read_all_bytes in vertex_imagen. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>	2026-03-12 13:22:56 -07:00
Ishaan Jaffer	add3183308	IGNORE_FUNCTIONS	2026-02-14 12:59:15 -08:00
Alexsander Hamir	ebce0e5f8c	[Release - 02/10/2026] v1.81.10-nightly	2026-02-10 16:26:30 -08:00
Krish Dholakia	7056d9984e	Custom Code Guardrails UI Playground (#20377) ... * feat(guardrails/): allow custom code execution for guardrails first step in allowing teams to submit custom code for guardrails * feat: custom_code_guardrail.md support passing custom code for guardrails * feat: initial commit adding ui for custom code guardrails allows users to write guardrails based on custom code * feat: expose new test custom code guardrail endpoint allows ui testing playground to sanity check if guardrail is working as expected * fix: fix linting errors * fix: fix max recursion check * fix: fix linting error	2026-02-03 19:57:24 -08:00
Ishaan Jaff	9ed11c5cdf	[Feat] Allow calling A2A agents through LiteLLM /chat/completions API (#20358) ... * init A2AConfig * add transform files * feat: A2A * feat A2AConfig * fix get_secret_str * init: A2AConfig * init A2AConfig common utils * A2AConfig * test_a2a_completion_async_non_streaming * fix * Update litellm/main.py Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com> * add multi part conversation support * extract_text_from_a2a_message --------- Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>	2026-02-03 12:52:33 -08:00
Ishaan Jaff	c23e4b87dc	[Feat] New LiteLLM Policy engine - create policies to manage guardrails, conditions - permissions per Key, Team (#19612) ... * init PolicyMatcher * TestPolicyMatcherGetMatchingPolicies * TestPolicyMatcherGetMatchingPolicies * feat: init PolicyResolver * init resolver types * init policy from config * inint PolicyValidator * validate policy * init Architecture Diagram * test_add_guardrails_from_policy_engine * init _init_policy_engine * test updates * test fixws * new attachment config * simplify types * TestPolicyResolverInheritance * fix policy resolver * fix policies * fix applied policy * docs fix * docs fix * fix linting + QA checks * fix linting + QA fixes * test fixes	2026-01-22 19:49:53 -08:00
Ishaan Jaffer	e3fe02148d	_mask_sequence	2026-01-10 13:20:43 -08:00
Alexsander Hamir	a7da4833da	[Fix] CI/CD - check_code_and_doc_quality (#18560)	2026-01-01 11:18:59 -08:00
Alexsander Hamir	892d7e8d70	[Fix] CI/CD - Fix Bedrock tool calling test failures with non-serializable objects and internal parameters (#17930) ... * fix(bedrock): filter non-serializable objects from request params - Enhanced filter_exceptions_from_params() to filter callable objects (functions) and Logging objects - Applied filtering in Bedrock's _prepare_request_params() before deepcopy - Applied filtering to additional_request_params before JSON serialization - Prevents TypeError during deepcopy (APIConnectionError objects) and JSON serialization (functions, Logging objects) - Fixes test_bedrock_tool_calling test failures Root cause: MCP-related functions (handle_chat_completion_with_mcp, completion_callable) and litellm_logging_obj were incorrectly added to optional_params via add_provider_specific_params_to_optional_params(), which then ended up in additional_request_params. These objects should be in litellm_params, not optional_params. * fix(bedrock): filter internal MCP parameters from API requests Filter out LiteLLM internal/MCP-related parameters (skip_mcp_handler, mcp_handler_context, _skip_mcp_handler) from additional_request_params before sending to Bedrock API to prevent 'extraneous key' errors. - Added filter_internal_params() helper function in core_helpers.py - Applied filtering in Bedrock's _prepare_request_params() method - Fixes test_bedrock_completion.py::test_bedrock_tool_calling * fix: mypy type error * fix: add filter_exceptions_from_params to recursive function ignore list - Add filter_exceptions_from_params to IGNORE_FUNCTIONS in recursive_detector.py - Function is safe: has max_depth parameter (default 20) to prevent infinite recursion	2025-12-13 12:38:07 -08:00
Sameer Kankute	b39c21d90c	fix: Add _delete_nested_value_custom to recursive function ignore list ... The _delete_nested_value_custom function is recursive but has bounded depth (limited by the number of path segments), preventing infinite recursion. This is necessary for nested field removal in additional_drop_params.	2025-12-09 17:56:13 +05:30
Ishaan Jaff	5c192a23c3	[Feat] Add new RAG API on LiteLLM AI Gateway (#17109) ... * init RAG api types * add RAG endpoints * init main.py for RAG ingest API * init RecursiveCharacterTextSplitter * add BaseRAGIngestion * fix OpenAIRAGIngestion * fix img handler * init OpenAIRAGIngestion * init BedrockRAGIngestion * init BedrockRAGIngestion * init rag tests * init BedrockVectorStoreOptions * implement BedrockRAGIngestion * add BaseRAGAPI * add endpoint for RAG ingest * add ingest RAG endpoints * add test doc * add parse_rag_ingest_request * update endpoints * docs add docs for new RAG API * fix qa check * fix linting * docs ficx * docs * add max depth checks * docs anthropic	2025-11-25 17:54:29 -08:00
yuneng-jiang	4b25398afe	[Infra] CI/CD Fixes (#16937) ... * Attempt CI/CD Fix * Adding test for coverage * Adding max depth to copilot and vertex * Fixing mypy lint and docker database * Fixing UI build issues * Update playwright test	2025-11-21 13:58:19 -08:00
Ishaan Jaffer	d32890ba55	fix _redact_base64	2025-10-28 17:38:16 -07:00
Ishaan Jaff	f1699d4cbf	fix IGNORE_FUNCTIONS	2025-09-08 17:33:36 -07:00
Ishaan Jaff	4c7b3e10ed	[Bug Fix] Gemini Tool Calling - fix gemini empty enum property (#14155) ... * fix: _convert_schema_types * fix recursive detector * test_convert_schema_types_type_array_conversion * fix: DEFAULT_NUM_WORKERS_LITELLM_PROXY * add _fix_enum_empty_strings * test_tool_call_with_empty_enum_property * test_fix_enum_empty_strings * fix _fix_enum_empty_strings	2025-09-01 17:05:01 -07:00
Ishaan Jaff	2331fb45d5	[Bug]: Gemini 2.5 Pro – schema validation fails with OpenAI-style type arrays in tools (#14154) ... * fix: _convert_schema_types * fix recursive detector * test_convert_schema_types_type_array_conversion * fix: DEFAULT_NUM_WORKERS_LITELLM_PROXY	2025-09-01 16:53:20 -07:00
Ishaan Jaff	621b3dca7b	[Bug Fix] Mistral Tool Calling - Grammar error: at 3(11): failed to compile JSON schema (#13389) ... * test_claude_tool_use_with_gemini * add _remove_json_schema_refs * add _clean_tool_schema_for_mistral * fixes mistral tool calls * _remove_json_schema_refs * fix - vertex, remove hardcoded test	2025-08-07 13:50:22 -07:00
Krish Dholakia	e7f1fa26ab	UI - Azure Content Guardrails (#12341) ... * build(model_prices_and_context_window.json): remove 'supports_tool_choice' for specific mistral models Closes https://github.com/BerriAI/litellm/issues/11750 * feat: initial commit adding cleaner ui for azure text moderation guardrails * feat(guardrail_endpoints.py): add discoverable guardrail configs and improve converting base model to dict with types * fix(guardrail_provider_fields.tsx): render from api endpoint correctly * fix(guardrail_provider_fields.tsx): cleanup * refactor(guardrail_endpoints.py): refactor to handle dictionaries with literal - allows multiselect * feat(ui/): render dictionary with known keys correctly * feat(ui/): render optional params on separate page * style(ui/): style improvements to rendering optional params on the UI * feat(azure/prompt_shield.py): add azure prompt shield back on UI * fix(add_guardrail_form.tsx): fix form to handle updated api * fix(guardrail_optional_params.tsx): ensure values are nested correctly for writing to api * fix: fix linting error * feat(text_moderation.py): handle str to int conversion * fix(guardrail_info.tsx): only render pii settings if guardrail is presidio * fix(guardrail_info.tsx): add guardrail specific fields to update settings allows updating guardrail fields (e.g. severity threshold) post-create * fix(guardrail_endpoints.py): set guardrail_id in guardrail object ensures duplicate objects not created on guardrail update * fix(guardrail_endpoints.py): allow provider specific fields to be updated on patch update * refactor(guardrail_endpoints.py): remove duplicate info endpoint * fix(guardrail_endpoints.py): mask sensitive keys on returning via guardrail `/info` Prevent leaking keys * fix(guardrail_optional_params.tsx): return numerical input when numerical component used fixes issue where output was a str * fix(guardrail_optional_params.tsx): render dict keys correctly * fix(text_moderation.py): fix severity by category check * fix(proxy/utils.py): check if guardrail should run for post call streaming hook Prevents invalid guardrails from running if not requested * test: fix import * fix: fix linting error * test: update test * fix: fix tests * fix: fix code qa errors * fix(guardrail_endpoints.py): set max depth for function * test: update recursive_detector.py * test: update list * build: merge main * fix: fix ruff check errors	2025-07-05 10:19:29 -07:00
Ishaan Jaff	f1c7024e70	[Feat] Add Bridge from generateContent <> /chat/completions (#12081) ... * add GenerateContentToCompletionHandler * working - non streaming bridge * add GoogleGenAIAdapter * add google gen ai adapter * working streaming bridge * working streaming usage for adapter * tool calling transform for generate content to openai * fixes for accumulating tool calls * fix code qa checks * Best Practices for Production * fix code qa checks * test_streaming_partial_tool_calls_accumulation * linting fixes * add supported_openai_chat_completion_params * fix translate_generate_content_to_completion * test_google_genai_adapter.py	2025-06-27 11:08:55 -07:00
Krish Dholakia	06484f6e5a	Xai, VertexAI, Google AI Studio - live web search support in OpenAI format (#11251) ... * build(model_prices_and_context_window.json): fix 'supports_web_search' flag - openai only supports it on 2 models - gpt-4o-search-preview and gpt-4o-mini-search-preview * feat(xai/chat): add xai web search options param support * test: add max tokens to test xai output very verbose * build(xai/): add web search support for all xai models * build(model_prices_and_cost.json): add gemini-2.0 supports web search * feat(gemini/): map openai 'web_search_options' to google's 'googlesearch' tool * build(model_prices_and_context_window.json): add supports_web_search for vertex_ai/gemini-2 models * fix: fix circular reference error * fix(convert_dict_to_response.py): handle scenario where xai returns finish reason as 'stop' for tool calls * fix: reduce function size * fix: import session handling * Revert "fix: import session handling" This reverts commit deb257dc10. * fix: linting pin mypy * [Feat]: Guardrails - Add streaming for bedrock post guard (#11247) * feat: add streaming for bedrock post guard * fix: bedrock guardrails * fix: add clear comments * Update litellm/proxy/guardrails/guardrail_hooks/bedrock_guardrails.py Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * Update litellm/proxy/guardrails/guardrail_hooks/bedrock_guardrails.py Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * fix: clean up bedrock guardrails --------- Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * [Fix] Responses API - Session management (#11254) * fix: import session handling * fix: imports for session handler * tests: tests for session handler * Update enterprise/litellm_enterprise/enterprise_callbacks/session_handler.py Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * bump: bump litellm enterprise * fixes: test_create_user_default_budget * fix(xai/): filter 'strict' on tool call * test: update test for new error string * fix(utils.py): default to None if not set in model cost map ensures consistent usage of 'supports_[x]' flags * fix(fireworks_ai/): support fireworks ai document inlining on pdf's sent via openai 'file' message type * test: update test * test: name filter_value_from_dict * fix(fireworks_ai/): handle cache control flag in messages * fix(xai/chat): fix check --------- Co-authored-by: Ishaan Jaff <ishaanjaffer0324@gmail.com> Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>	2025-05-31 14:26:16 -07:00
Krish Dholakia	fdfef04d93	Gemini Multimodal Live API support (#10841) ... * fix: initial commit * refactor(gemini/realtime/transformation.py): initial instrumentation * fix: fix default api base if not set * feat(gemini/): passes initial user request to backend * feat(realtime_streaming.py): support transforming message input before sending it enables gemini realtime streaming to be sent in correct format * feat: initial working commit of setup message being sent + working * fix(gemini/): initial commit supporting realtime response transformation * feat(gemini/realtime): transform session.created event correctly * fix(realtime_streaming.py): more gemini/realtime response mapping - handle new message * test(gemini/realtime/test_): add more unit tests * feat(gemini/realtime): handles consecutive deltas * feat(gemini/realtime): support openai 'response.text.done' event * feat(gemini/realtime): add openai 'response.text.done' and 'response.content_part.done' event support * fix(gemini/realtime): add openai 'response.done' event support unified realtime api support * fix: fix linting errors * fix: fix linting error * fix: fix linting error * fix: handle infinite loop * fix: fix linting error * fix: fix recursive detector * fix: fix file * fix: fix linting error * fix: fix linting error	2025-05-15 22:39:09 -07:00
Krish Dholakia	7210b713dc	Add target model name validation (#10722) ... * fix(auth_checks.py): enforce auth checks on target model names ensures user has access to models they are trying to call * test(test_auth_utils.py): add unit tests for auth check * fix(exception_mapping_utils.py): handle mistral 429 exception * fix: fix linting error * fix(auth_checks.py): add max fallback depth	2025-05-10 14:27:06 -07:00
Krish Dholakia	bf9382a182	Handle more gemini tool calling edge cases + support bedrock 'stable-image-core' (#10351) ... * test(test_amazing_vertex_completion.py): try to repro https://github.com/BerriAI/litellm/issues/10319 * fix(common_utils.py): handle edge case on tools Fixes https://github.com/BerriAI/litellm/issues/10319 * test: add unit testing for infinite loops * fix(amazon_stability3_transformation.py): support 'stable-image-core' transformation Fixes https://github.com/BerriAI/litellm/issues/8488 * test: add unit testing for stable image core model * test: update test	2025-04-28 14:22:29 -07:00
Krish Dholakia	fdfa1108a6	Add property ordering for vertex ai schema (#9828) + Fix combining multiple tool calls (#10040) ... * fix #9783: Retain schema field ordering for google gemini and vertex (#9828) * test: update test * refactor(groq.py): initial commit migrating groq to base_llm_http_handler * fix(streaming_chunk_builder_utils.py): fix how tool content is combined Fixes https://github.com/BerriAI/litellm/issues/10034 * fix(vertex_ai/common_utils.py): prevent infinite loop in helper function * fix(groq/chat/transformation.py): handle groq streaming errors correctly * fix(groq/chat/transformation.py): handle max_retries --------- Co-authored-by: Adrian Lyjak <adrian@chatmeter.com>	2025-04-15 22:29:25 -07:00
Krish Dholakia	87733c8193	Fix anthropic prompt caching cost calc + trim logged message in db (#9838) ... * fix(spend_tracking_utils.py): prevent logging entire mp4 files to db Fixes https://github.com/BerriAI/litellm/issues/9732 * fix(anthropic/chat/transformation.py): Fix double counting cache creation input tokens Fixes https://github.com/BerriAI/litellm/issues/9812 * refactor(anthropic/chat/transformation.py): refactor streaming to use same usage calculation block as non-streaming reduce errors * fix(bedrock/chat/converse_transformation.py): don't increment prompt tokens with cache_creation_input_tokens * build: remove redisvl from requirements.txt (temporary) * fix(spend_tracking_utils.py): handle circular references * test: update code cov test * test: update test	2025-04-09 21:26:43 -07:00
Krish Dholakia	8e3c7b2de0	fix(vertex_ai.py): move to only passing in accepted keys by vertex ai response schema (#8992) ... * fix(vertex_ai.py): common_utils.py move to only passing in accepted keys by vertex ai prevent json schema compatible keys like $id, and $comment from causing vertex ai openapi calls to fail * fix(test_vertex.py): add testing to ensure only accepted schema params passed in * fix(common_utils.py): fix linting error * test: update test * test: accept function	2025-04-07 18:07:01 -07:00
Nicholas Grabar	1f2bbda11d	Add recursion depth to convert_anyof_null_to_nullable, constants.py. Fix recursive_detector.py raise error state	2025-03-28 13:11:19 -07:00
Ishaan Jaff	fff15543d9	(UI + Proxy) Cache Health Check Page - Cleanup/Improvements (#8665) ... * fixes for redis cache ping serialization * fix cache ping check * fix cache health check ui * working error details on ui * ui expand / collapse error * move cache health check to diff file * fix displaying error from cache health check * ui allow copying errors * ui cache health fixes * show redis details * clean up cache health page * ui polish fixes * fix error handling on cache health page * fix redis_cache_params on cache ping response * error handling * cache health ping response * fx error response from cache ping * parsedLitellmParams * fix cache health check * fix cache health page * cache safely handle json dumps issues * test caching routes * test_primitive_types * fix caching routes * litellm_mapped_tests * fix pytest-mock * fix _serialize * fix linting on safe dumps * test_default_max_depth * pip install "pytest-mock==3.12.0" * litellm_mapped_tests_coverage * add readme on new litellm test dir	2025-02-19 19:08:50 -08:00
Krish Dholakia	f651d51f26	Litellm dev 02 07 2025 p2 (#8377) ... * fix(caching_routes.py): mask redis password on `/cache/ping` route * fix(caching_routes.py): fix linting erro * fix(caching_routes.py): fix linting error on caching routes * fix: fix test - ignore mask_dict - has a breakpoint * fix(azure.py): add timeout param + elapsed time in azure timeout error * fix(http_handler.py): add elapsed time to http timeout request makes it easier to debug how long request took before failing	2025-02-07 17:30:38 -08:00
Ishaan Jaff	53a3ea3d06	(Refactor) Langfuse - remove prepare_metadata, langfuse python SDK now handles non-json serializable objects (#7925) ... * test_langfuse_logging_completion_with_langfuse_metadata * fix litellm - remove prepare metadata * test_langfuse_logging_with_non_serializable_metadata * detailed e2e langfuse metadata tests * clean up langfuse logging * fix langfuse * remove unused imports * fix code qa checks * fix _prepare_metadata	2025-01-22 22:11:40 -08:00
Ishaan Jaff	dd385410df	(Code quality) - Ban recursive functions in codebase (#7910) ... * code qa add RecursiveFunctionFinder * test_recursive_detector * RecursiveFunctionFinder * fix check * recursive_detector	2025-01-21 20:33:32 -08:00