😩 Why MCPHub?

If you work with AI tools in 2026, you probably have multiple MCP servers — filesystem, GitHub, databases, APIs. Managing them is painful:

• ❌ Scattered configs — every AI client needs its own MCP server list
• ❌ No visibility — which servers are running? What tools are available?
• ❌ No single endpoint — each client connects to each server separately
• ❌ Restart hell — add a server = restart everything

MCPHub fixes all of this:

• ✅ One YAML file to define all your MCP servers
• ✅ One gateway endpoint that routes to the right server
• ✅ Beautiful dashboard showing status, tools, and stats
• ✅ Hot reload — edit config, servers update instantly

🏗️ Architecture

┌─────────────────────────────────────────────────┐
│                   AI Clients                     │
│         (Claude, Cursor, Copilot, etc.)          │
└──────────────────┬──────────────────────────────┘
                   │  MCP Protocol (stdio)
                   ▼
┌─────────────────────────────────────────────────┐
│              🔱 MCPHub Gateway                   │
│                                                  │
│   ┌──────────┐ ┌──────────┐ ┌───────────────┐  │
│   │ Registry │ │ Router   │ │  Health Check  │  │
│   └──────────┘ └──────────┘ └───────────────┘  │
│                                                  │
│   ┌──────────────────────────────────────────┐  │
│   │        🌐 Web Dashboard (:3000)          │  │
│   └──────────────────────────────────────────┘  │
└──────┬──────────────┬──────────────┬────────────┘
       │              │              │
       ▼              ▼              ▼
┌────────────┐ ┌────────────┐ ┌────────────┐
│ filesystem │ │   github   │ │  postgres  │
│   server   │ │   server   │ │   server   │
└────────────┘ └────────────┘ └────────────┘

🚀 Quick Start

1. Install

npm install -g mcphub

2. Create mcphub.yaml

mcphub:
  port: 3000

servers:
  - name: filesystem
    command: npx
    args: ["-y", "@modelcontextprotocol/server-filesystem", "/home/user"]

  - name: github
    command: npx
    args: ["-y", "@modelcontextprotocol/server-github"]
    env:
      GITHUB_TOKEN: "${GITHUB_TOKEN}"

3. Run

mcphub

Open http://localhost:3000 to see the dashboard 🎉

✨ Features

📖 Config Reference

mcphub:
  port: 3000            # Dashboard port (default: 3000)
  log_level: info       # debug | info | warn | error

servers:
  - name: my-server     # Unique server name (required)
    command: npx         # Command to start the server (required)
    args: ["-y", "pkg"]  # Command arguments (optional)
    enabled: true        # Enable/disable without removing (optional)
    env:                 # Environment variables (optional)
      KEY: "value"
      SECRET: "${FROM_ENV}"  # Reference host env vars

Environment Variable Substitution

Use ${VAR_NAME} in any string value to reference the host environment:

env:
  GITHUB_TOKEN: "${GITHUB_TOKEN}"     # From host env
  DATABASE_URL: "${DATABASE_URL}"

🔧 Using with AI Clients

Configure your AI client to connect to MCPHub instead of individual servers:

{
  "mcpServers": {
    "mcphub": {
      "command": "mcphub",
      "args": ["/path/to/mcphub.yaml"]
    }
  }
}

All tools from all configured servers are now available through this single connection.

🤝 Contributing

Contributions are welcome! Here's how:

1. Fork the repo
2. Create your feature branch: git checkout -b feat/amazing-feature
3. Commit: git commit -m 'Add amazing feature'
4. Push: git push origin feat/amazing-feature
5. Open a Pull Request

Development

git clone https://github.com/NeuZhou/mcphub.git
cd mcphub
npm install
npm run dev

📄 License

MIT © Kang Zhou