---
title: My Stars Atlas
publishedAt: "2026-07-20"
summary: A generated, searchable catalog of starred GitHub repositories grouped into
  stable categories, built with a deterministic classification engine and
  published as a static site on GitHub Pages.
tags: [TypeScript, GitHub API, GitHub Pages, GitHub Actions, Handlebars, Static Site, Awesome List, Automation]
language: "en"
type: "project"
status: "published"
---

# My Stars Atlas

A generated catalog of starred GitHub repositories, grouped into stable
categories.

[Homepage](https://sametcn99.github.io/my-stars-atlas/) |
[GitHub Repository](https://github.com/sametcn99/my-stars-atlas)

## About This Project

This repository generates a curated, searchable catalog from a GitHub account's
starred repositories. It combines a data pipeline, deterministic categorization
rules, chunked JSON exports, and a static frontend so the result can be
published on GitHub Pages without a server runtime.

The project is useful if you want to:

- publish your starred repositories as a clean public directory,
- keep repository metadata and categories in version control,
- customize how projects are grouped with your own rules and overrides,
- reuse the structure for your own profile by forking or cloning the repository.

## How It Works

The generator fetches starred repositories for `@sametcn99`, classifies each
repository with stable rule-based categories, writes the results into chunked
files under `dist/data/`, and rebuilds the README together with the static site
assets. The frontend then reads the generated catalog and chunk files to provide
filtering, search, sorting, and progressive loading in the browser.

The main project-level settings live in `config/config.json`, while
classification behavior stays in `config/categories.json` and
`config/overrides.json`.

## Key Features

- **Deterministic Classification Engine**: The taxonomy lives in configuration,
  not hardcoded per repository. Overrides win first, then rule-based scoring
  tries to find the strongest category. Only low-confidence items may trigger
  extra README fetches, which keeps the fallback precise and avoids paying that
  cost for the full dataset.
- **Chunked JSON Exports**: Catalog data is split into chunked files under
  `dist/data/` for progressive loading in the browser.
- **Static Frontend**: Filtering, search, sorting, and progressive loading —
  all client-side, no server runtime required.
- **Automated Updates**: A GitHub Actions workflow keeps the catalog in sync
  with newly starred repositories on a schedule.
- **Forkable & Reusable**: Fork the repo, edit `config/config.json` for your
  own username, add a GitHub token, and run the generator on your fork.

## Use This Project

You can use this project in three common ways:

1. **Fork It** — Keep the existing setup and adapt it for your own GitHub
   account. Edit `config/config.json` for your own username, README copy, and
   site SEO settings.
2. **Clone It** — Get full local control or customize the data model,
   categorization rules, UI, or deployment flow.
3. **Contribute To It** — Improve category definitions, UI and accessibility,
   GitHub API efficiency, or documentation.

## Technology Stack

- **TypeScript**
- **Bun**
- **Handlebars** (templating)
- **GitHub Actions** (automation)
- **GitHub Pages** (hosting)