See any public repo as one cross-layer map — your code, your database tables, your API routes, and your product rules, joined. Pick a node and watch the blast radius ripple across every layer.
Most graphs map code to code. This links code ↔ DB tables (from your Prisma/SQL schema) ↔ API routes ↔ business rules (from your PRD) — the join no single-layer tool draws.
Every node and edge is derived from a real file. Nothing is guessed or hallucinated — a business rule with no code anchor stays honestly unknown.
Your repo is shallow-cloned to a temp dir, scanned, and deleted on the spot. Nothing persists on the server. For private repos, run it locally.