Copilot Service API

This Flask application provides an API service for AI-powered code completions, mimicking GitHub Copilot functionality.

Features

• Generate code completions with configurable AI models
• Token usage tracking
• Prompt customization
• Caching for efficiency
• Interactive playground for testing

PyPI Installation

pip install copilot-service

GitHub Installation

pip install git+https://github.com/yourusername/copilot-service.git

Usage

Set API Keys

The service uses LiteLLM for model access. To declare API keys, use LiteLLM's standard format. For example:

export OPENAI_API_KEY=your-api-key-here
# Or for other providers
export COHERE_API_KEY=your-api-key-here
export ANTHROPIC_API_KEY=your-api-key-here

Start the Service

copilot-service

Configuration

You can also configure the default model and token limits:

export DEFAULT_MODEL="gpt-3.5-turbo"
export DEFAULT_MAX_TOKENS=100
copilot-service

API Endpoints

POST /copilot/completions

Get AI-generated code completions

Request Body:

{
  "file_contents": "def hello_w",
  "cursor_position": 10
}

Response:

{
  "completion": "orld():\n    print(\"Hello World!\")",
  "token_metadata": {
    "prompt_tokens": 120,
    "completion_tokens": 20,
    "total_tokens": 140
  }
}

POST /copilot/preview

Preview the prompt being sent to the AI

GET /version

Get service version information

Testing

Run tests with:

python test.py

License

GNU General Public License v3.0

Interactive Playground

Access the playground at http://localhost:5000

• Type code and click "Get Completion" for ghost text suggestions
• Test API endpoints directly from the browser