Stack
Next.js, FastAPI, Python, TypeScript, Three.js, OpenAI
Background
Onboarding to a new codebase is one of the slowest parts of joining a project. Reading through file trees and chasing import chains to build a mental model takes hours. GitBoarding came from wanting a faster way to understand how a codebase is structured at a glance, and to be able to ask questions about it in plain language rather than grep-ing through files.
Overview
GitBoarding takes any GitHub repository and transforms it into an interactive 3D dependency graph. It parses imports across the entire codebase to build a directed file graph, then runs GraphSAGE embeddings and Louvain community detection to group related files into semantically named modules. Node sizes scale with PageRank, so the most critical files in the repo immediately stand out visually. Paired with the graph is a chat interface where you can ask natural language questions about the codebase, with relevant clusters retrieved through semantic search, actual source files are read in context, and conversation history is preserved so follow-up questions feel natural.