RepoLens — AI-Powered Open-Source Repository Intelligence

RepoLens is an AI-powered platform designed to help developers understand unfamiliar GitHub repositories, explore project documentation, and identify where to begin their open-source contribution journey.

The platform uses Retrieval-Augmented Generation (RAG) to retrieve relevant repository information and generate context-aware answers supported by source citations. It offers configurable retrieval strategies, including hybrid search combining vector similarity and BM25 keyword search using Reciprocal Rank Fusion (RRF), enabling better retrieval of both semantically related information and exact technical terms.

Powered by a locally running Gemma 4 model through Ollama, RepoLens aims to provide repository-aware assistance while reducing dependence on hosted LLM APIs for core inference. Its planned capabilities include relevance-based abstention when evidence is insufficient, a structured “Where Do I Start?” onboarding checklist, and experiments comparing RAG versus non-RAG responses, retrieval configurations, and answer quality.

Built with Next.js, Node.js, Express, LangChain.js, LangGraph.js, Qdrant, and MongoDB, RepoLens aims to bridge the gap between discovering an open-source project and making a meaningful first contribution. Its central goal is to make repository knowledge more accessible, answers more verifiable, and open-source onboarding more actionable.