1k GitHub Stars

Interactive treemap of 60k+ GitHub repositories with daily momentum, curated discovery, and language-level exploration.

Live site: https://ustars.dev

Overview

Search example

What it does

• Projects tab: default repo map, with curated and guide-style repositories filtered out
• Daily tab: highlights momentum using star-growth deltas from the latest snapshot
• Awesome tab: brings back the filtered awesome / guide / tutorial / interview style repositories as a dedicated view
• Language drill-down: click any language block to zoom into that language only
• Hover metadata: repo created/updated dates are loaded from a static index instead of per-hover API calls
• Search: matches repository name, description, and language labels across the active view

Data model

The app is driven by a single snapshot file:

• data/repos.json — main repo dataset
• public/repo-meta.json — lightweight hover metadata index

Each repo row currently stores:

1. fullName
2. stars
3. forks
4. langIdx
5. description
6. growth
7. createdAt
8. updatedAt

Refreshing GitHub data

Use the refresh script:

npm run refresh:growth

What it does:

• refreshes stars / forks / timestamps from GitHub
• computes growth against the previous local snapshot
• writes a resumable progress file during long runs
• snapshots the previous dataset into data/snapshots/
• regenerates public/repo-meta.json

Local development

npm install
npm run dev

Open:

• http://localhost:3000

Build

npm run build

This project is configured as a static export.

Cloudflare Pages deployment

The current deployment path is Cloudflare Pages:

npm run build
npx wrangler pages deploy out --project-name github-treemap --branch main

Repository structure

app/                  Next.js App Router pages
components/           treemap canvas, header, tooltip, panel
data/                 repo snapshot data
lib/                  grouping, filtering, classifier, metrics
public/repo-meta.json static hover metadata index
scripts/              data refresh scripts

Notes

• The project intentionally separates the main treemap data from hover metadata to reduce runtime latency and avoid per-hover server/API requests.
• Curated repositories are filtered at the data layer, not only in the UI, so homepage / language pages / daily view stay consistent.