GitHub Stars Explorer

Interactive, static GitHub Pages app for exploring a GitHub starred-repository collection with search, category filters, presets, CSV export, and a responsive force-directed graph.

Live Demo · Update Workflow

Current Dataset

• Total repositories: 1,574 in the checked-in dataset
• Last synced: May 5, 2026 19:23 UTC
• Source: GitHub Actions generated JSON
• Live GitHub count at validation time: 1,574, which GitHub rounds as 1.6k
• Deployment target: GitHub Pages

The app shows the exact checked-in dataset count and marks data as stale if the sync timestamp is older than 24 hours.

Features

• D3 force-directed graph with category-aware clustering and cluster labels
• Default Sample view with a small balanced set to keep first load easy to scan, especially on phones
• Presets for Sample, All, Curated, Popular, Learning, and AI
• Real-time search by name, owner, description, topics, and language
• Compact Top Matches panel for scannable search results
• Category, language, and star-count filters
• Exact count display, sync status, last synced date, and stale-data warning
• CSV export of repository names, URLs, categories, secondary categories, topics, stars, forks, and sync timestamp
• Mobile-responsive panels and touch-friendly graph interactions
• GitHub Pages friendly: static HTML, CSS, JavaScript, and JSON only

Category Model

The generated data keeps the original category field for compatibility and adds a more external-user-friendly display taxonomy:

Display Category	Count
AI Agents	263
Frontend	211
Docs/Reference	189
Backend/API	128
AI Models	105
AI Learning	82
Other	78
Python	77
Developer Tools	77
RAG/Search	71
Full-stack Frameworks	44
Career/Interview	41
LLM Apps	38
Databases	34
Systems/Low-level	33

Repositories can also include secondaryCategories so visitors can understand cross-cutting projects without breaking older consumers of data/repositories.json.

Project Structure

starred-repos-graph/
├── index.html
├── app.js
├── data/
│   └── repositories.json
├── scripts/
│   ├── fetch-data.js
│   ├── clean-data.js
│   └── validate-data.js
├── tests/
│   └── e2e/
└── .github/workflows/
    └── update-data.yml

Local Development

python3 -m http.server 8080

Then open http://localhost:8080/.

Run validation:

node --check app.js
node --check scripts/fetch-data.js
node --check scripts/validate-data.js
node scripts/validate-data.js

Run browser tests:

cd tests
npm install
npm test -- --project=chromium

Updating The Dataset

The scheduled GitHub Actions workflow fetches starred repositories, categorizes them, validates the JSON, commits data changes, and deploys the static app to GitHub Pages.

For a different account, set GITHUB_USERNAME in the workflow or run locally:

GITHUB_USERNAME=your-github-username node scripts/fetch-data.js > data/repositories.json
node scripts/validate-data.js

The app can be configured from the small config object in app.js for username, repository URL, workflow URL, and stale-data threshold.

CI/CD Safety

The validation script checks:

• JSON shape and required repository fields
• totalCount against the actual repository array length
• dataset age and stale-data threshold
• category distribution and Other ratio
• secondary category presence
• live GitHub starred count within a small tolerance

This keeps the project static and GitHub Pages compatible while making count mismatches visible before deployment.