GitBoarding

Codebase onboarding tool that helps developers understand unfamiliar GitHub repositories through an interactive 3D dependency graph and a RAG-based chat interface.

What it does

3D dependency graph - parses imports across the whole repo to build a directed file graph.

Repo chat - ask anything about the codebase. Every file is embedded once per repo and stored in a per-repo ChromaDB collection; each question is embedded and matched against it via cosine similarity, pulling in only the files that are actually relevant

Stack

Frontend	Next.js, TypeScript, Tailwind
3D graph	3d-force-graph (Three.js)
Inference	FastAPI
Vector store	ChromaDB (one collection per repo)
Embeddings	OpenAI text-embedding-3-small
LLM	GPT-4o-mini (streaming)

Running locally

Inference server

cd inference
pip install -r requirements.txt
uvicorn main:app --reload --port 8000

Frontend

cd web
npm install
npm run dev

Env vars needed:

.env (root):

GITHUB_TOKEN=...
OPENAI_API_KEY=...

web/.env.local:

INFERENCE_URL=http://localhost:8000