A GitHub profile analyzer built by prompting an AI, not by writing every line myself.
A few weeks ago I decided to try something I'd only heard about in passing: vibe coding — building a real, working product mostly by describing what I wanted to an AI coding agent, rather than writing every line of code by hand.
I didn't come from a heavy engineering background, and honestly, that was the point. I wanted to see how far "clear thinking + clear prompting" could actually take a product from idea to something live and usable.
The result is GitGlow — a tool that turns any public GitHub profile into a visual dashboard.
🔗 Try it here: https://git-glow-three.vercel.app/
What GitGlow Actually Does
You paste in a GitHub username (or a full profile URL), hit analyze, and it pulls together:
- 🔥 Contribution streaks — current streak and longest streak
- 💻 Language breakdown — a donut chart of what languages you actually write, weighted by real code volume, not just repo count
- 📈 Language trend over time — how your stack has evolved year by year
- 🕐 Best time to code — a commit-hour heatmap that tells you if you're a night owl, an early bird, a strict 9-to-5 coder, or a weekend warrior
- ✅ Repo health score — a 0–100 score per repo based on license, description quality, recent activity, and issue tracking
- 🏢 Personal vs. organization contributions — how much of your work is solo vs. through teams/orgs
- 🆚 Compare mode — put two GitHub profiles side by side and see who's really grinding harder
- 🎴 Shareable dev card — a generated image you can drop into your README or share on social
- 🧠 AI-written summary — a short, natural-language paragraph describing your coding personality, generated live from your stats
Everything is fetched live from GitHub's API on every request. Nothing about you is stored — no database, no accounts, no tracking. You paste a username, you get a dashboard, that's it.
Why I Built This
I wanted a project that was genuinely useful, not just a toy demo — something developers would actually enjoy poking around in. GitHub profiles already have so much interesting data sitting in them: when you code, what you code in, how consistent you are, how healthy your repos are. Nobody was surfacing it in one visual place, so I built the thing I wanted to use myself.
There's also a slightly competitive, fun layer to it — the compare feature exists because half the fun of a tool like this is checking your stats against a friend's, a colleague's, or your favorite open-source maintainer's.
The Vibe Coding Experience
This was my first time building something end-to-end this way, and a few things stood out:
1. The bottleneck moves from "how do I write this" to "what exactly do I want."
Once you're not fighting syntax, the real work becomes describing the feature precisely enough that the output actually matches your intent. Vague prompts get vague results — being specific about data flow, edge cases, and UI behavior mattered far more than I expected.
2. Iteration speed changes how you design.
Because building a feature took minutes instead of hours, I found myself trying things I normally wouldn't have bothered prototyping — like the commit-hour heatmap or the shareable dev card. Low cost of experimentation meant more experiments.
3. You still need to think like an engineer.
Vibe coding doesn't remove the need to understand your product. I still had to reason about API rate limits, what happens when a token isn't configured, what a "repo health score" should even mean, and how to keep the whole thing stateless by design. The AI handled implementation; the decisions were still mine.
4. Shipping beats perfecting.
It's tempting to keep adding "one more feature." At some point I just pushed it live, and that turned out to be the right call — real usage and feedback teach you more than another week of solo polishing ever could.
Tech Stack
- Next.js (App Router) + TypeScript
- Tailwind CSS + shadcn/ui
- Recharts for the charts and heatmaps
- GitHub REST + GraphQL APIs for all the data
- Gemini API for the AI-generated summary (optional — falls back to a template if unset)
- Deployed on Vercel
No backend database. No user accounts. Everything server-rendered on demand from live GitHub data.
What's Next
A few things I'm considering adding:
- Full contribution calendar (the classic GitHub-style green squares grid)
- Pinned repos support
- An opt-in public leaderboard (this one actually needs a small database, unlike everything else here — deliberately holding off until it's worth the tradeoff)
- Better error/empty states and loading skeletons
Try It and Tell Me What You Think
This is genuinely my first project built this way, and I'd love honest feedback — what's confusing, what's missing, what you'd want to see next. If you try it, paste your own GitHub username in and see if the "night owl or 9-to-5 coder" label actually gets you right.
🔗 https://git-glow-three.vercel.app/
If you build something similar or have ideas for what to add, drop a comment — I'm actively iterating on this.
If you found this useful, a clap or a share helps more people find it — and if you're curious about vibe coding yourself, happy to answer questions in the comments.
Top comments (0)