Where is tool calling actually executed?

[jbkoh]

Hi there, thanks for the great project!

I'm trying to debug how LibreChat is using available tools. I'm particularly interested in what's passed to OpenAI model as input to create tool arguments. I'm using a slightly old version of the code, but I'd like to know which part of the current code base is responsible for that.

The workflow I have is choose an MCP, and based on the chat history I see an MCP tool is called with some arguments. I want to know how/where the arguments are decided within the code.

Thanks a lot in advance!