Get the FREE Ultimate OpenClaw Setup Guide →

mcp-gateway

MCP Gateway and Registry

Installation
Run this command in your terminal to add the MCP server to Claude Code.
Run in terminal:
Command
claude mcp add --transport stdio aarora79-mcp-gateway docker run -i aarora79/mcp-gateway:latest \
  --env LOG_LEVEL="info|debug|warn|error" \
  --env REGISTRY_URL="URL of the MCP Registry (if applicable)" \
  --env NGINX_HOSTNAME="Hostname for the gateway (if reverse proxy is externalized)" \
  --env MCP_GATEWAY_CONFIG="Path to gateway/registry config or inline configuration"

How to use

MCP Gateway & Registry provides a unified entry point to access multiple MCP servers. It acts as a reverse proxy and discovery hub, letting AI Agents connect to various MCP-enabled services through a single endpoint and enabling governance, health monitoring, and dynamic tool discovery. The registry portion surfaces available MCP servers and their tools, while the gateway handles routing, authentication, and health checks. To use it, deploy the gateway (and registry if you need discovery UI) behind a reverse proxy or load balancer, point your agents to https://your-gateway-domain/{server-name}/, and let the gateway route requests to the appropriate MCP backend. Tools advertised by registered MCP servers will be discoverable by the UI or programmatic APIs, and agents can request tool lists, schemas, and execution endpoints via the standard MCP protocol (SSE or Streamable HTTP).

How to install

Prerequisites:

  • Docker installed on your host (or a container orchestration platform-compatible runtime).
  • Access to a network where the gateway can be exposed to AI Agents.

Installation steps:

  1. Pull the gateway image (or build from source if you have the repository): docker pull aarora79/mcp-gateway:latest

  2. Run the gateway with a basic configuration: docker run -d --name mcp-gateway
    -p 443:443
    -e MCP_GATEWAY_CONFIG=/path/to/config.json
    -e REGISTRY_URL=https://registry.example.com
    aarora79/mcp-gateway:latest

  3. If you have an internal registry, ensure the MCP server definitions are registered with the gateway (via UI/API or through a configuration file loaded at startup).

  4. Optional: Run a separate registry container if you want the discovery UI and management features. Follow the same docker run pattern and configure it to point at the gateway.

  5. Verify installation by curling the gateway health endpoint or visiting the UI (if enabled).

Notes:

  • Replace image names, URLs, and config paths with your actual values.
  • If using Kubernetes, translate the docker run into a Deployment/StatefulSet with proper ConfigMaps and Secrets.

Additional notes

Tips and common considerations:

  • Ensure TLS certificates are properly configured if exposing the gateway to the internet. Use a reverse proxy (Nginx or similar) in front of the gateway for TLS termination.
  • Use the Registry UI/API to register and manage MCP servers; this enables dynamic discovery for agents.
  • The gateway supports SSE and Streamable HTTP transports; ensure your MCP servers implement compatible endpoints.
  • Define health checks for registered MCP servers to keep the registry’s status up to date. Health metrics can be exposed via the gateway’s web UI or API.
  • Environment variables can be used to tune logging, enable authentication, and configure registry endpoints. Keep sensitive values in a secrets store and inject at runtime.
  • If you need to scale, run multiple gateway instances behind a load balancer and use a shared registry state or a backing store for service registration.

Related MCP Servers

Sponsor this space

Reach thousands of developers