CodebaseAI

An AI-powered codebase assistant using Retrieval-Augmented Generation (RAG) to understand your repositories. Upload any GitHub repository, and chat with it to understand its architecture, find where authentication is handled, or locate dead code.

Key Features

• Semantic Code Search: Uses vector embeddings and ChromaDB.
• RAG Architecture: Answers questions using context from your own code.
• Source Citations: Highlights file names and line numbers.
• Local LLMs: Powered by Ollama.
• FastAPI Backend: Robust REST API for repository indexing and querying.
• Premium React Frontend: Sleek glassmorphism UI built without node.js.

Quickstart

1. Clone this repository.
2. Install dependencies:

pip install -r requirements.txt

3. Start an Ollama server locally:

ollama run llama3

4. Run the application:

python main.py

5. Open http://localhost:8000/ in your browser.