Mcp Rest Adapter

MCP → REST Adapter

Make your existing REST APIs instantly accessible to AI agents (Claude, GPT-4, etc.) without changing a single line of code.

🚀 Quick Start (5 minutes)

# Clone the repository
git clone https://github.com/deepwissen/mcp_rest_adapter.git
cd mcp_rest_adapter

# Start everything with one command
./quick_start.sh

That's it! Your REST APIs are now AI-accessible. Test it:

curl -X POST http://localhost:8000/mcp \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc": "2.0", "method": "tools/list", "params": {}, "id": 1}'

🎯 What This Does

The MCP Adapter acts as a bridge between AI agents and your existing microservices:

AI Agent (Claude/GPT) → MCP Protocol → MCP Adapter → Your REST APIs

Before: Months of custom AI integration code After: 5 minutes to AI-enable your entire API ecosystem

🛠️ Manual Setup

Prerequisites

• Docker & Docker Compose
• Python 3.11+
• Your REST APIs with OpenAPI/Swagger specs

Step 1: Configure Your Services

Edit docker-compose.yml to point to your services:

environment:
  - CUSTOMER_SERVICE_URL=http://your-customer-api:8080
  - ORDER_SERVICE_URL=http://your-order-api:8080
  - INVENTORY_SERVICE_URL=http://your-inventory-api:8080

Step 2: Start the Adapter

# Start with Docker Compose
docker-compose up -d

# Or run directly with Python
pip install -r requirements.txt
python -m uvicorn mcp_adapter.server:app --port 8000

Step 3: Verify It's Working

# Check health
curl http://localhost:8000/health

# List available tools (auto-generated from your APIs)
curl -X POST http://localhost:8000/mcp \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc": "2.0", "method": "tools/list", "params": {}, "id": 1}'

📋 Requirements for Your APIs

Your REST APIs need:

1. Health endpoint (/health) that returns 200 OK
2. OpenAPI spec (/openapi.json or /swagger.json)
3. Standard REST patterns (GET, POST, PUT, DELETE)

That's it. No code changes required.

🔧 Configuration

Create a config.py file:

SERVICE_CONFIGS = {
    "customer": {
        "base_url": "http://customer-service:8001",
        "health_endpoint": "/health",
        "openapi_endpoint": "/openapi.json"
    },
    "order": {
        "base_url": "http://order-service:8002",
        "health_endpoint": "/health",
        "openapi_endpoint": "/openapi.json"
    }
}

🧪 Testing

Run the test suite:

# Run all tests
./run_tests.py

# Run with Docker
docker-compose -f docker-compose.test.yml up

📊 What You Get

Once running, AI agents can:

• Discover all your API endpoints automatically
• Call any endpoint with proper parameters
• Handle complex workflows across multiple services
• Get structured responses in MCP format

Example: An AI agent can now say "Show me all orders for customer John Doe" and automatically:

1. Search for the customer
2. Get their customer ID
3. Fetch all orders
4. Return formatted results

🚨 Troubleshooting

Services not discovered?

• Check your services have /health endpoints returning 200 OK
• Verify /openapi.json is accessible

No tools generated?

• Ensure your OpenAPI spec is valid
• Check logs: docker-compose logs mcp-adapter

Connection errors?

• Verify network connectivity between adapter and your services
• Check service URLs in configuration

🤝 Contributing

We welcome contributions! Please check out our [Contributing Guide](CONTRIBUTING.md).

📝 License

MIT License - see [LICENSE](LICENSE) file for details.

🔗 Links

• Documentation
• Issues
• Discussions

---

Built with ❤️ to make AI integration simple

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: deepwissen
• Source: deepwissen/mcprest_adapter
• License: MIT

Install and usage instructions live in the source repository linked above.