Techiral/NLP-Commander

NLP Command Processor

A FastAPI-based backend system that processes natural language commands into structured actions for SaaS services.

Features

• Natural language command processing
• Modular adapter architecture for service integration
• API key authentication
• Command status tracking
• Extensible design for adding new service adapters

Project Structure

backend/
├── app/
│   ├── adapters/         # Service adapters
│   ├── models/          # Data models
│   ├── services/        # Core services
│   └── main.py         # FastAPI application
├── tests/              # Unit tests
└── requirements.txt    # Python dependencies

Setup

1. Create a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

2. Install dependencies:

   pip install -r requirements.txt

3. Run the server:

   uvicorn app.main:app --reload

API Documentation

Once the server is running, visit:

• OpenAPI documentation: http://localhost:8000/docs
• ReDoc documentation: http://localhost:8000/redoc

Authentication

All endpoints require an API key header:

X-API-Key: your-secret-key

Endpoints

POST /process-command

Process a natural language command.

Request:

{
  "command": "Schedule a team meeting"
}

Response:

{
  "command_id": "20250401100000",
  "status": "completed",
  "processed_intent": {
    "service": "calendar",
    "action": "create_event",
    "parameters": {
      "title": "Team Meeting",
      "date": "2025-04-01",
      "time": "10:00 AM"
    }
  }
}

GET /status/{command_id}

Get the status of a processed command.

Extending the Adapter Layer

To add a new service adapter:

1. Create a new adapter class in app/adapters/ that inherits from BaseAdapter
2. Implement the required methods:
   • service_name property
   • execute() method
3. Register the adapter in MCPService initialization

Example:

class EmailAdapter(BaseAdapter):
    @property
    def service_name(self) -> str:
        return "email"
    
    def execute(self, intent: Dict[str, Any]) -> bool:
        # Implement email sending logic
        return True

Running Tests

pytest