Added MCP support for chat Completion endpoint.

[Simon-Lind-glitch]

Title

Added MCP execution for chat completion endpoint

Relevant issues

Fixes bug #14268

Pre-Submission checklist

Please complete all items before asking a LiteLLM maintainer to review your PR

• I have Added testing in the tests/litellm/ directory, Adding at least 1 test is a hard requirement - see details
• I have added a screenshot of my new test passing locally
• My PR passes all unit tests on make test-unit
• [/] My PR's scope is as isolated as possible, it only solves 1 specific problem

Type

🐛 Bug Fix

Changes

Ive added support for MCP tool exeuction to the chat completions pipeline.
While I was at it I also fixed the tool format in the requests towards bedrock since it was the model i was testing with.
Reasonably related issues.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Preview	Comments	Updated (UTC)
litellm	Ready	Preview	Comment	Sep 16, 2025 6:35am

[krrishdholakia]

why is this specific to bedrock? @Simon-Lind-glitch

[Simon-Lind-glitch]

We've only gottent his error while using bedrock and did not find any bug regarding this, so we thought we'd isolate this to bedrock

[krrishdholakia]

the issue your closing refers to ollama and all non-openai providers

the fix would be incorrect to apply if it's just for bedrock. Can we please have this be a more generic fix across providers?

[krrishdholakia]

modifying the request body should happen via a async_pre_call_hook - this is cleaner than modifying the routing logic - https://docs.litellm.ai/docs/proxy/call_hooks

see -

litellm/litellm/proxy/hooks/cache_control_check.py

Line 19 in 8e22cf5

async def async_pre_call_hook(

[Simon-Lind-glitch]

Ill get on it

[Simon-Lind-glitch]

i need to tinker with this a bit more. ill get back to you once its ready