This project is a full-stack application that uses an AI agent to autonomously generate documentation for a given GitHub repository. It features a React-based frontend and a Python backend powered by LangGraph, with a focus on real-time user engagement through WebSockets and efficient parallel processing.
The application is designed with a clear separation of concerns between the frontend, backend, and the AI agent core.
-
Frontend: A modern Vite + React single-page application provides the user interface. Users can submit a GitHub repository URL and watch the documentation process unfold in real-time. A WebSocket connection streams live logs and status updates, creating an engaging and interactive experience.
-
Backend: A Python-based backend built with FastAPI serves as the main hub. It manages a Celery task queue with a Redis broker to handle long-running documentation jobs. This architecture allows for parallel processing of repository files, significantly reducing the time required to generate documentation.
-
AI Core: The heart of the project is an "AI Agent" built with LangGraph. The agents performs the following steps:
- Clones the target GitHub repository.
- Recursively analyzes the file structure.
- Dispatches file analysis tasks to the Celery queue for parallel execution.
- Synthesizes the results to generate a complete set of Markdown documentation, including per-folder
README.mdfiles, function-level explanations, and a "how to run" guide.
-
Data & Infrastructure: Generated documentation is stored, and a link is provided to the user. The combination of Celery and Redis ensures that the system is scalable and can handle multiple documentation requests concurrently.
-
Submission: A user pastes a GitHub URL into the React frontend.
-
Job Creation: The frontend sends the URL to the backend via a WebSocket, which initiates a new documentation job.
-
Parallel Processing: The AI agent clones the repo and adds file analysis tasks to the Celery queue. Multiple workers pick up these tasks and process files in parallel.
-
Real-time Feedback: Throughout this process, the agent and workers stream logs and progress updates back to the frontend over the WebSocket. The UI displays these updates in a live terminal.
-
Completion: Once all analysis is complete, the agent assembles the final documentation and sends it to the frontend. The user can then view and copy the generated Markdown.
| Area | Technology | Purpose |
|---|---|---|
| Backend | Python, FastAPI | API endpoints and WebSocket management |
| AI Core | LangGraph | Orchestrating the autonomous documentation agent |
| Queue | Celery, Redis | Parallel processing and handling long-running jobs |
| Frontend | React, Vite, Tailwind CSS | Modern, responsive user interface |
| Real-time | WebSockets | Live communication between frontend and backend |
| Data | S3 / Local Storage | Storing generated documentation |
This combination of technologies creates a powerful, efficient, and user-friendly tool for automated code documentation.