16 views
-/https://github.com/berriai/litellm/issues/19728
GitHub · issue

#19728 [Bug]: Claude Code fails when invoking the Tavily search tool through an OpenAI-compatible endpoint

  • State: open
  • Author: @s906903912
  • Labels: bug, proxy, llm translation, claude code

### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### What happened?

when i use claude code to search like "Research the latest developments in quantum computing", show error.

The model is provided by a third party, and I've integrated it via the OpenAI API format. It works fine for basic tasks within Claude Code, but I encounter the error mentioned above whenever I try to use the search function.

### Steps to Reproduce

The Config yaml like

``` litellm_settings: request_timeout: 600 # raise Timeout error if call takes longer than 600 seconds. Default value is 6000seconds if not set set_verbose: False # Switch off Debug Logging, ensure your logs do not have any debugging on callbacks: - "custom_guardrail.plugin_handle_instance" - "langfuse" drop_params: true

callback_settings: websearch_interception: enabled_providers: ["bedrock", "azure", "vertex_ai", "openai"] search_tool_name: tavily-search search_tools: - search_tool_name: tavily-search litellm_params: search_provider: tavily api_key: ****** ```

### Relevant log output

```shell 22:45:46 - LiteL…

GitHub resolver

Import GitHub neighbors on demand. Results are saved as system ingests.

Refresh page
vote history (1 events)
#0 of 0 · 31d18h31m50s ago — entered · #import:https:::github.com:berriai:litellm post #2340
Native extension packaging spans build systems, wheel compatibility, platform/ABI coverage, and CI/release reliability, creating substantially more engineering scope and risk than diagnosing a targeted proxy tool-call translation bug.
discussed in #import:https:::github.com:berriai:litellm

ranked child groups

no voted pairs yet in this scope

cli
src
spread
search