lightfast-mcp

---
description: Guidelines for adding new MCP servers and extending the lightfast-mcp project
globs: 
alwaysApply: false
---

# Server Extension Guide

## Adding New MCP Servers

### 1. Create Server Directory Structure
```
src/lightfast_mcp/servers/{app}/
├── __init__.py
├── server.py          # Main server implementation
├── tools.py           # Tool implementations
└── resources.py       # Resource implementations (optional)
```

### 2. Create Entry Point Script
Create `src/lightfast_mcp/servers/{app}_mcp_server.py`:

@src/lightfast_mcp/servers/mock_server.py

Follow this template pattern, replacing `mock` with your application name.

### 3. Add to pyproject.toml
Add new script entry:
```toml
[project.scripts]
lightfast-{app}-server = "lightfast_mcp.servers.{app}_mcp_server:main"
```

### 4. Server Auto-Discovery
New servers placed in `src/lightfast_mcp/servers/{app}/` following the `server.py` pattern (containing a class that inherits from `BaseServer` and defines `SERVER_TYPE`) are automatically discovered by the `ServerRegistry` located in `src/tools/orchestration/server_registry.py`. No manual changes to the registry code are typically needed.

### 5. Add Configuration Support
Update sample configurations to include the new server type

## Server Implementation Pattern
```python
from ..core.base_server import BaseServer
from fastmcp import mcp

class {App}MCPServer(BaseServer):
    """MCP server for {App} integration."""
    
    def __init__(self, config):
        super().__init__(config)
        self.setup_{app}_connection()
    
    @mcp.tool()
    async def {app}_tool(self, param: str) -> str:
        """Tool description for AI."""
        # Implementation
        return "result"
```

## Testing New Servers
1. Create unit tests in `tests/unit/test_{app}_server.py`
2. Add integration tests in `tests/integration/test_{app}_integration.py`
3. Include E2E tests if the application supports automation
4. Test both stdio and HTTP transports