Back to work

Featured

gitGood

Full-Stack / AI Workflow Developer

AI code analysis tool that converts GitHub repos into interactive, beginner-friendly walkthroughs — it maps structure, generates diagrams, and creates step-by-step tutorials using Mastra and LLM-assisted analysis.

Overview

The problem

Open-source repos can be overwhelming. Outdated docs and disorganized files create a barrier to contribution. GitGood simplifies this by converting any GitHub repo into a clear, interactive tutorial.

The solution

GitGood analyzes a repo's files, architecture, and logic, generates high-level maps and diagrams, and produces multi-page, beginner-friendly walkthroughs with clickable node trees and explanations.

What it does

  • Fetches a GitHub repo and extracts code structure automatically.
  • Generates a high-level overview and maps component relationships.
  • Breaks down code into digestible tutorial chapters with diagrams.
  • Provides interactive, clickable code diagrams for exploration.

How we built it

Tech: React + TypeScript (Next.js) frontend, Mastra Agent workflows, Google Gemini, libSQL, and a backend that fetches repos, runs analysis, and packages a UI-ready JSON payload for the frontend.

Challenges

Integrating frontend and backend workflows, avoiding AI hallucination by enforcing structured prompts and file splitting, and building a robust pipeline that outputs consistent, UI-ready data.

Accomplishments & what's next

We learned Mastra end-to-end, shipped a working prototype, and plan to add language support, a built-in AI chatbot per page, and smarter, skill-adaptive explanations.

Stack
Next.jsTypeScriptMastraAI AgentsMarkdown