Claude Code's built-in self-knowledge?
Problem: I find conflicting information regarding what Claude Code has knowledge of in regards to its own tools and configurations. It seems to know some stuff, but then it has to retrieve documents from docs.anthropic over every tool or question I ask it about its built in stuff. It has no idea that it's Claude Code, even though it's in it's status, in doctor, in its system prompt.. so it has no idea even given the specific URL to mcp documentation that it needs the claude code mcp configuration not the claude desktop configuration, and even after telling it, it still doesn't find .mcp.json which I believe is the configuration for it. Also, having a hell of a time getting claude code properly setup in windows with wsl. is it even possible to have it setup correctly without warnings and errors in windows 11? because claude and all the documentation and tutorial videos has failed to fix the issues and its been months of trying
This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗