mcpwatch

MCPWatch

Know what your AI agents can access. MCPWatch is a small command-line tool that statically audits local Model Context Protocol (MCP) configuration files and flags risky settings before you connect them to an AI coding agent. It also keeps local snapshots so you can see exactly what changed in a configuration over time. MCPWatch only reads the configuration file you point it at. It never starts your MCP servers, never runs the commands it finds, and never sends data over a network. Github Repository: https://github.com/hasheramin5-cyber/MCPWatch