MCP Gateway Registry

Clone the MCP Gateway & Registry repository from GitHub to your local machine or server.

git clone https://github.com/agentic-community/mcp-gateway-registry.git
cd mcp-gateway-registry

Copy the environment template and edit it with your credentials. At minimum set KEYCLOAK_ADMIN_PASSWORD, REGISTRY_API_TOKEN, and DOCUMENTDB_HOST (or MONGODB_CONNECTION_STRING for Atlas).

cp .env.example .env
# Edit .env with your values:
# KEYCLOAK_ADMIN_PASSWORD=your-admin-password
# REGISTRY_API_TOKEN=your-static-api-token
# DOCUMENTDB_HOST=localhost
# AUTH_PROVIDER=keycloak
# EMBEDDINGS_PROVIDER=sentence-transformers

Launch the full stack using pre-built images from Docker Hub. The gateway will be available at http://localhost:7860. On non-Apple Silicon systems use the --prebuilt flag; Apple Silicon requires --podman.

export DOCKERHUB_ORG=mcpgateway
./build_and_run.sh --prebuilt

Point your AI coding assistant or agent framework at the gateway's unified endpoint. All tool discovery, authentication, and routing is handled by the gateway.

{
  "mcpServers": {
    "gateway": {
      "command": "npx",
      "args": ["mcp-gateway-registry"],
      "env": {
        "REGISTRY_URL": "http://localhost:7860",
        "REGISTRY_API_TOKEN": "your-static-api-token"
      }
    }
  }
}

Use the REST API or the included CLI to register your backend MCP servers and AI agents with the registry. Once registered, they are discoverable via semantic or lexical search.

# Register an MCP server
curl -X POST http://localhost:7860/api/servers \
  -H "Authorization: Bearer your-static-api-token" \
  -H "Content-Type: application/json" \
  -d '{"id": "my-db-server", "transport": "streamable-http", "url": "http://my-db-server:3000"}'

Run the included tests to confirm the deployment is working correctly. The test suite covers unit, integration, and end-to-end scenarios.

make test-unit