{"id":"metamcp","name":"MetaMCP","homepage":"https://github.com/metatool-ai/metamcp","repo_url":"https://github.com/metatool-ai/metamcp","category":"gateway","subcategories":["mcp-gateway","aggregator","middleware","orchestrator"],"tags":["gateway","aggregator","middleware","docker","postgresql","sse","streamable-http","openapi","rate-limiting","multi-tenancy","oauth","api-key","pnpm","turbo"],"what_it_does":"MetaMCP is a self-hosted MCP aggregator and gateway that groups multiple MCP servers into namespaces, exposes them as unified endpoints with configurable auth and rate limiting, and supports tool customization, middleware, and OpenAPI output in a single Docker deployment.","use_cases":["Dynamically composing and remixing tools from multiple MCP servers into named endpoints consumed by Claude Desktop or Cursor","Applying middleware transformations (e.g., filtering inactive tools) to optimize token usage when serving LLMs","Multi-tenant deployments where different users or teams each get isolated namespaces with their own API keys and tool selections"],"not_for":["Users who need a simple single-server setup without aggregation","Teams without Docker or Node.js operational capabilities","Use cases requiring an on-premise install without any external network calls (PostgreSQL and Docker required)"],"best_when":"You need a full-featured self-hosted MCP gateway with namespace grouping, per-endpoint rate limiting, multi-tenancy, tool overrides, and OpenAPI compatibility in one Docker Compose deployment.","avoid_when":"You want a lightweight single-binary gateway, a managed cloud service, or your needs are covered by connecting one MCP server directly to one client.","alternatives":["mcpjungle","mesh","mcpo"],"af_score":74.3,"security_score":72.0,"reliability_score":null,"package_type":"mcp_server","discovery_source":["github"],"priority":"low","status":"evaluated","version_evaluated":"latest","last_evaluated":"2026-03-01T09:50:05.944007+00:00","performance":{"latency_p50_ms":null,"latency_p99_ms":null,"uptime_sla_percent":null,"rate_limits":"Configurable per endpoint and per user via token bucket algorithm","data_source":"llm_estimated","measured_on":null}}