This project, Codebase RAG, implements an AI expert over a codebase using Retrieval-Augmented Generation (RAG). The web app allows users to chat with a codebase to understand how it works and identify areas for improvement.
- Chat with Codebase: Users input queries, and the most relevant code snippets are retrieved to generate responses using an LLM.
- Embedding Codebase: The contents of the codebase are embedded and stored in a vector database called Pinecone for efficient retrieval.
- Web App: A user-friendly interface for interacting with the codebase.
- Multiple Codebases: Allow users to select different codebases to chat with.
- Webhook Integration: Update the Pinecone index automatically when new commits are pushed to the repository.
say you will find the direction for the front-end and the back-end in each folder inside
- Example Web App: Link
- Blog: RAG on Codebases Blog
- Article: Getting Started with Embeddings
- Embedding Model Leaderboard: Link
- How Greptile does Codebase RAG: Article
- RAG for a Codebase with 10k Repos: Article
- How Embeddings are Generated: Paper
This project is licensed under the MIT License - see the LICENSE file for details.