Auth MCP Server

Fast, low-latency MCP server for authentication services Optimized for real-time AI agent integration with LM Studio, Claude Desktop, and more.

🚀 Quick Start

# 1. Clone the repository
git clone https://github.com/wattlesol/auth-mcp.git
cd auth-mcp

# 2. Install dependencies
npm install

# 3. Configure environment
cp .env.example .env
# Edit .env with your API URLs

# 4. Test the server
./test-mcp-stdio.sh

# 5. Configure your AI agent (see Configuration section below)

✨ Features

• ⚡ Low Latency - Direct stdio communication (< 10ms)
• 🔧 18 Dynamic Tools - Auto-loaded from Swagger/OpenAPI spec
• 🤖 AI Agent Ready - Works with LM Studio, Claude Desktop, etc.
• 🐳 Docker Support - Optional containerized deployment
• 📚 Complete Docs - Comprehensive guides and examples

📋 Configuration

For LM Studio

{
  "mcpServers": {
    "auth-mcp": {
      "command": "node",
      "args": ["/absolute/path/to/auth-mcp/auth-mcp"],
      "env": {
        "AUTH_API_BASE_URL": "https://your-auth-api.example.com",
        "SWAGGER_URL": "https://your-auth-api.example.com/api-json"
      }
    }
  }
}

For Claude Desktop

Same configuration as above. Add to:

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

For Qwen Coding Agent

Configuration is in .qwen/settings.json (see QWEN-SETUP.md for details)

🎯 Available Tools (18 Total)

The server dynamically loads tools from your Swagger/OpenAPI specification. Example tools:

Authentication & Sign In:

• post_api_auth_signin - Sign in with email/password
• post_api_auth_staff_signin - Staff sign in (branch-aware)
• post_api_auth_signup - Create new user account
• post_api_auth_signout - Sign out current user

User Info & Profile:

• get_api_auth_me - Get current user information
• put_api_auth_profile - Update user profile

OTP & Verification:

• post_api_auth_otp_send - Send OTP to email
• post_api_auth_otp_verify - Verify OTP code
• post_api_auth_verify_email - Verify email with OTP
• post_api_auth_verify_phone - Verify phone with OTP

Password Management:

• post_api_auth_forgot_request - Request password reset
• post_api_auth_change_password - Change password

Token & Health:

• post_api_auth_refresh_token - Refresh JWT token
• get_api_health - Health check endpoint

📚 Documentation

🧪 Testing

# Run automated tests
./test-mcp-stdio.sh

# Manual test
echo '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"test","version":"1.0.0"}}}' | node auth-mcp

🏗️ Architecture

AI Agent (LM Studio/Claude)
         ↓ stdio (JSON-RPC)
Auth MCP Server (Node.js)
         ↓ HTTPS
Backend Auth API (Swagger)

📊 Performance

• Latency: < 10ms (stdio communication)
• Startup: ~2 seconds (loads Swagger spec)
• Memory: ~50MB (Node.js + dependencies)
• Protocol: MCP 2024-11-05

🐳 Docker Deployment

# Build
docker build -f Dockerfile.mcp -t auth-mcp-stdio .

# Run
docker run --rm -i --env-file .env auth-mcp-stdio

🔧 Environment Variables

Copy .env.example to .env and configure:

AUTH_API_BASE_URL=https://your-auth-api.example.com
SWAGGER_URL=https://your-auth-api.example.com/api-json
AUTH_API_TIMEOUT=5000

🛠️ Development

# Install dependencies
npm install

# Run HTTP server (legacy)
npm start

# Run MCP server (stdio)
npm run mcp

# Run tests
npm test

📦 Project Structure

├── auth-mcp                      # Main launcher script
├── src/
│   ├── mcp-stdio-server.js       # MCP server (stdio)
│   ├── api-client.js             # Auth API client
│   ├── swagger-client.js         # Swagger API client
│   └── swagger-tools-analyzer.js # Tool generator
├── config/
│   ├── lm-studio-config.json     # LM Studio config
│   ├── claude-desktop-config.json # Claude Desktop config
│   └── mcp-config-docker.json    # Docker config
├── docs/                         # Documentation
├── test/                         # Test files
├── .env.example                  # Example environment variables
└── package.json                  # Dependencies

🤝 Contributing

This is a production-ready MCP server. Feel free to:

• Add more tools
• Improve error handling
• Add monitoring/metrics
• Extend documentation

📄 License

ISC

🆘 Support & Troubleshooting

Issues? Check these guides:

1. QWEN-SETUP.md - Qwen Coding Agent setup & troubleshooting
2. QWEN-TOOLS-FIX.md - Tools not showing? Read this
3. docs/QUICK-SETUP.md - Quick troubleshooting
4. docs/CONFIGURATION-GUIDE.md - Detailed help

Enable Debug Mode:

MCP_DEBUG=true node auth-mcp

🎉 Success!

You now have a production-ready MCP server for AI agents!

Next Steps:

1. Configure your AI agent with the settings above
2. Test with: "Can you list the available authentication tools?"
3. Start using the tools in your AI conversations!