* fix: resolve recursion in OVHCloud get_supported_openai_params (#24111)
Root cause: OVHCloudChatConfig.get_supported_openai_params() called
get_model_info() which called back into get_supported_openai_params(),
causing infinite recursion and falling back to no tools support.
Made-with: Cursor
* fix: use .get() for TypedDict + don't hardcode function_calling support
Made-with: Cursor
---------
Co-authored-by: themavik <themavik@users.noreply.github.com>
Added a new section to the config.yaml documentation explaining how to
set the LITELLM_LICENSE environment variable for enterprise features.
Co-authored-by: Krish Dholakia <krrishdholakia@gmail.com>
- Skip short-circuit for providers that have a BaseAnthropicMessagesConfig
(bedrock, vertex_ai, azure_ai, anthropic) — they use the agentic loop
which includes a follow-up LLM synthesis step. Short-circuiting would
return raw search text instead of an LLM-synthesized answer.
- Add fallback to litellm.get_llm_provider() for custom_llm_provider
derivation when litellm_params is overwritten by kwargs.
- Add test for bedrock guard.
Addresses Greptile review comments #3 and #4.
Tests were outdated after _get_and_validate_existing_key was refactored
to use prisma_client.db.litellm_verificationtoken.find_unique() and
ProxyException. Also add ProxyException handling in bulk_update_keys
error extractor so error messages aren't empty.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Tests were outdated after _get_and_validate_existing_key was refactored
to use prisma_client.db.litellm_verificationtoken.find_unique() instead
of prisma_client.get_data(), and to raise ProxyException instead of
HTTPException. Also fix bulk_update_keys error handler to extract
ProxyException.message (str(ProxyException) returns empty string).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add None guard for prisma_client before calling update_data, and add
"unblocked" to AUDIT_ACTIONS literal type.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Split the single try/except in the _exception_raised cleanup path into
separate try blocks for asyncio.create_task and executor.submit, matching
the pattern used in _run_deferred_stream_guardrails. If create_task
raises, sync logging via executor.submit still fires.
Replace sys.exc_info()[1] check with an explicit _exception_raised boolean
sentinel. The flag is function-scoped, immune to outer exception context,
and only set when an exception actually occurs in base_process_llm_request.
This prevents false positives when called from a caller's except block.
- Replace hand-rolled _extract_search_query with existing
get_last_user_message from common_utils
- Use full UUID (str(uuid.uuid4())) to match codebase convention
- Move uuid import to module level per CLAUDE.md
