Navigating a new repository can be overwhelming. I built "Codebase Intelligence" tool to turn static code into an interactive knowledge base using Retrieval-Augmented Generation. Instead of the AI guessing what your code does, it reads the relevant files before answering.
By using semantic search and vector embeddings, you can ask questions like:
"How is the authentication flow handled?"
"Where are the API routes defined?"
Get a context-aware answer backed by your actual code.
I reached some key milestones while building this tool: automated an ingestion pipeline using LangChain and OpenAI embedding model to fetch, chunk, and embed GitHub repos. Leveraged Pinecone vector database for high-performance semantic search and metadata filtering. Integrated GPT 4.0 and Vercel AI SDK to manage the conversation flow. Implemented GitHub Actions to handle automated daily maintenance and cleanup of the database.
Check it out here: https://codebase-intelligence-nu.vercel.app/
Open Source and Contributions 🌟
I've made this tool open source! Whether you want to use it for your own repos or help improve the ingestion logic, feel free to check out the code or create an Issue.
Github Repository Link: https://github.com/nancy-kataria/codebase-intelligence
Top comments (0)