How can cursorLens log the interactions with LLM in Agent mode? #38
Description
In Cursor agent mode, cursor sends messages (prompt + tools, e.g., functions/mcp tools) to LLM, and receives the tools to be called and some messages, then it can run tools and do other interactions with LLM.
I guess CursorLens can only log the 1st round messages to LLM and unable to know the following interactions in agent mode because it is unknow how cursor internally uses the LLM result and feed it back to LLM. Is that right?