AutoPR-AI: Intelligent Code Review API

AutoPR-AI is a backend REST API that enables automated code review for files. It leverages Github OAuth, OpenAI GPT-4, and MongoDB Atlas to analyze code and leave intelligent review comments directly on pull requests.

• GitHub OAuth Login
• GPT-4 Powered Code Review
• MongoDB Atlas for Data Persistence
• JWT Authentication
• RESTful API Design using FastAPI

Features

• 📁 Upload a file via /review/upload
• 🤖 Generate a GPT-based review via /review/analyze
• 🔀 Create GitHub Pull Requests via /github/pr
• 💬 Add automated review comments via /github/comment
• 🔐 Authenticate users via GitHub OAuth via /oauth/login
• 👉 Issue JWT tokens and manage user sessions

Tech Stack

• FastAPI – High-performance Python API framework
• MongoDB Atlas + Motor – Asynchronous database access
• GitHub OAuth + PyGithub – GitHub authentication & PR actions
• OpenAI GPT-4 API – AI-powered code analysis and comment generation
• JWT (pyjwt) – Secure user session management
• Swagger (OpenAPI) – Auto-generated documentation at /docs

Authentication Flow

• Visit /oauth/login to authenticate via GitHub
• Receive a jwt_token on successful login
• Pass the token in headers for protected endpoints: Authorization: Bearer <jwt_token>

Configuration

Environment variables are required. See docs/api-usage.md for full setup.

Getting Started

1. Clone the repository

   git clone https://github.com/yourusername/AutoPR-AI.git
   cd AutoPR-AI

2. Set up the virtual environment

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

3. Install dependencies

    pip install -r requirements.txt

4. Configure environment variables
   • Create a .env file with the following keys:

GITHUB_TOKEN=your_github_pat
GITHUB_CLIENT_ID=your_client_id
GITHUB_CLIENT_SECRET=your_client_secret
OPENAI_API_KEY=your_openai_api_key

5. Run the server

uvicorn app.main:app --reload

API Reference

Detailed endpoint descriptions and request/response examples are available in API_DOC.md.

(or) Auto-generated Swagger UI is available at: http://localhost:8000/docs

Future Improvements

• Optional cloud deployment via GCP or Azure
• Smarter GPT-4 prompt engineering
• GitHub webhook support for automatic reviews
• Frontend dashboard (React/Next.js) as a future extension

License

This project is licensed under the MIT License.