azerdsq131/mcpm

mcpm is a command-line tool that manages MCP servers across the AI coding tools on your machine. MCP, or Model Context Protocol, is the system that gives AI assistants like Claude, Cursor, or VS Code Copilot access to external capabilities: reading your files, querying databases, calling APIs, and more. Without a manager, adding or removing an MCP server means opening configuration files by hand for each tool separately. mcpm handles that for you. You install or remove a server with a single command, and mcpm automatically writes the correct configuration format for every AI client it detects on your system. It supports nine clients out of the box, including Claude Code, Cursor, VS Code Copilot, Windsurf, Zed, Gemini CLI, and several others. Servers can run as Node.js packages, Python tools, Docker containers, Go binaries, or Deno scripts, and mcpm knows how to launch each type correctly. The registry bundled with mcpm lists over 1,000 verified MCP servers. You can search it from the command line and install curated bundles grouped by theme, such as webdev, data, AI integrations, DevOps, or cloud provider APIs. A doctor command checks your installed servers for broken configurations and reports on their health. For teams, mcpm supports a project-level configuration file called .mcpmrc where you list the servers your project needs. A teammate clones the repo, runs mcpm sync, and their AI clients are configured to match. Before any write operation, mcpm saves a snapshot of the current config so you can roll back if something breaks. The package is published on npm as mcp-fleet. The license is MIT.