GitMap

Inspiration

Searching GitHub can be overwhelming. The built-in search expects you to know the right filters and keywords, but most developers just think in plain language. We wanted a tool that lets you search GitHub the way you think, not the way you type. That idea inspired GitMap.

What it does

GitMap is an AI-powered GitHub search agent. A user types a natural language query like "instagram clone in javascript" and GitMap uses Gemini to generate the right search keywords and filters. Those keywords are then fed into the GitHub API to return the most relevant repositories. The result: a more intuitive way to discover projects, libraries, and tools.

How we built it

Frontend: React + TailwindCSS for a clean, responsive interface.
Backend: Node.js + Express to handle API calls.
AI: Google Gemini API to transform natural language queries into structured GitHub searches.
GitHub API: Fetches repository data (name, stars, license, last update) using the generated keywords.

Challenges we ran into

Rate limits: GitHub’s API restricts unauthenticated requests, so we had to be careful with testing and caching results.
First-time setup: Setting up the backend, environment variables, and connecting Gemini was new to some of us.

Accomplishments that we're proud of

Building a working end-to-end pipeline: natural language → Gemini → GitHub API → clean UI results.
Creating a tool that feels genuinely useful for developers who struggle with GitHub’s advanced search syntax.

What we learned

How to integrate a generative AI model into a real product workflow.
How important prompt design and output filtering are when relying on AI for structured results.
Best practices with environment variables, Express APIs, and handling rate limits.