Spaces:
Running
Running
metadata
title: Codebase Explainer
emoji: π
colorFrom: blue
colorTo: green
sdk: docker
app_file: app.py
pinned: false
π AI Codebase Explainer
Ask questions about any GitHub repository in plain English. No more reading 50 files manually.
π Live Demo
π Try it here
π Benchmark Results (psf/requests)
| Metric | Result |
|---|---|
| Files indexed | 46 files |
| Chunks created | 341 chunks |
| Ingestion time | 15.45 seconds |
| Vector DB latency | ~100ms |
| Accuracy | 4/5 questions β |
π οΈ Tech Stack
- LangChain + RAG pipeline
- ChromaDB (vector store)
- HuggingFace Embeddings (all-MiniLM-L6-v2)
- Groq LLaMA 3 (LLM)
- Streamlit (UI)
- Docker + HuggingFace Spaces (deployment)
π‘ How It Works
- Paste any GitHub repo URL
- App clones and indexes all code files
- Ask questions in plain English
- Get answers with source file citations
π― Use Cases
- Understand a new codebase in minutes
- Onboard to a new company faster
- Explore open source projects instantly