A lightweight deployment configuration for MetaMCP with custom dependencies.

This repository contains a minimal setup for deploying MetaMCP with custom dependencies:

• Dockerfile: Extends the official MetaMCP image (ghcr.io/metatool-ai/metamcp:latest) with additional tools
• docker-compose.yml: Production deployment configuration
• GitHub Actions: Automated builds and pushes to ghcr.io/drejom/metamcp:latest

This setup adds the following to the base MetaMCP image:

• Tailscale CLI for secure networking
• PostgreSQL client tools for database operations

The core MetaMCP functionality comes from the official upstream image, ensuring you automatically receive updates and improvements.

1. Copy example.env to .env and configure your environment variables:

   cp example.env .env

2. Deploy with Docker Compose:

   docker-compose up -d

GitHub Actions automatically:

• Builds the custom Dockerfile when you push to main or create version tags
• Pushes the resulting image to ghcr.io/drejom/metamcp:latest
• Can be deployed via Dokploy or any container orchestration system

This lightweight approach provides several benefits over maintaining a full fork:

• Automatic upstream updates: Base MetaMCP improvements are inherited automatically
• Minimal maintenance: Only need to manage your specific customizations
• Fast builds: Only installing additional dependencies, not rebuilding the entire application
• Clear separation: Your customizations vs upstream MetaMCP code

MetaMCP is a powerful MCP proxy that aggregates multiple MCP servers into a unified interface. For complete documentation, features, and usage instructions, visit:

• Documentation: https://docs.metamcp.com
• Official Repository: https://github.com/metatool-ai/metamcp
• Discord Community: https://discord.gg/mNsyat7mFX

Claude Code stores MCP server configurations in ~/.claude.json. You can configure MetaMCP in several ways:

1. Global configuration (recommended - available across all projects): Edit ~/.claude.json and add to the mcpServers section:

   {
     "mcpServers": {
       "metamcp": {
         "type": "stdio",
         "command": "npx",
         "args": ["@drejom/metamcp", "http://your-metamcp-url:port", "YOUR_API_KEY"]
       }
     }
   }

2. Project-level configuration (shared with team): Create a .mcp.json file in your project root:

   {
     "mcpServers": {
       "metamcp": {
         "command": "npx",
         "args": ["@drejom/metamcp", "http://your-metamcp-url:port", "YOUR_API_KEY"]
       }
     }
   }

3. Using Claude Code CLI:

   # Add MetaMCP server globally (user scope)
   claude mcp add metamcp --scope user -- npx @drejom/metamcp http://your-metamcp-url:port YOUR_API_KEY
   
   # Or for project scope (creates .mcp.json)
   claude mcp add metamcp --scope project -- npx @drejom/metamcp http://your-metamcp-url:port YOUR_API_KEY

Note: MetaMCP should use stdio transport, not HTTP transport, for proper tool exposure.

For Claude Desktop, edit the configuration file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
Linux: ~/.config/Claude/claude_desktop_config.json

Add MetaMCP to the mcpServers section:

{
  "mcpServers": {
    "metamcp": {
      "command": "npx",
      "args": ["@drejom/metamcp", "http://your-metamcp-url:port", "YOUR_API_KEY"]
    }
  }
}

For secure API key management:

• Claude Code: Use --env flag or add to .mcp.json
• Claude Desktop: Set in your shell profile or use the env field in config

MIT - Same as the upstream MetaMCP project.