I think we need more organization principles around what tools are given to the LLM/AI. Ideally, organizing it by roles, purpose and even layering on access control. We're working on a solution for this at MintMCP

→ View original post on X — @jiquanngiam,