MetaMCP

by metatool-ai

MetaMCP is a single Docker container that aggregates, orchestrates, and proxies all your MCP servers behind one unified gateway. If you are running 5, 10, or 20 separate MCP servers across different projects, MetaMCP replaces the chaos of managing individual server configs with a centralized dashboard. The core problem it solves: MCP servers multiply fast. You add one for GitHub, one for your database, one for file access, one for Slack, another for web search. Each needs its own configuration in Claude, Cursor, or whatever client you use. MetaMCP sits in front of all of them. Your client connects to one endpoint. MetaMCP routes tool calls to the right server behind the scenes. Key architecture decisions that set MetaMCP apart: Namespace-based grouping: Organize MCP servers by project, team, or function. A "frontend-dev" namespace might expose GitHub, Figma, and Vercel servers. A "data-eng" namespace exposes BigQuery, Postgres, and Airflow. Tool remixing: Cherry-pick specific tools from different servers into custom endpoints. You do not have to expose every tool from every server. Pluggable middleware: Add observability (logging, tracing) and security (auth, rate limiting) as middleware layers without modifying server code. Dual transport: Supports both SSE and Streamable HTTP. Public endpoints get API key + OAuth authentication via OpenID Connect. 2.2K GitHub stars with 317 forks. 797 commits. MIT licensed. Self-hosted via Docker Compose — no cloud dependency. For teams looking to build AI workflows on top of aggregated MCP servers, see Langflow for visual pipeline building. For governance of the agents consuming these tools, check OpenBox AI.

FAQ