🪦 GitHub Graveyard

Enter a GitHub username, get a tombstone for every abandoned side project, each with a cause of death.

RIP todo-app-v3 · 2022 – 2023 · Cause of death: Replaced by its own sequel

How it works

• A grave is any repo the user owns (not a fork, not the profile README) that is archived or hasn't had a push in 180+ days.
• Causes of death (lib/epitaphs.ts) are deterministic, so the same repo always dies the same way. Name/description jokes come first (todo apps, -v2 rewrites, AI wrappers, crypto, tutorials, Advent of Code…), then trait jokes (archived, famous, lots of open issues, died in 2 days, language). Each cause is used at most once per graveyard.
• Ranks ("Weekend Warrior", "Serial Abandoner", "Mass Grave Operator"…) come from the dead-to-alive ratio and the average lifespan.
• Data comes from GitHub's REST API and is cached for hours with "use cache" + cacheLife("hours"). Repos are fetched least-recently-pushed first (up to 300), so even prolific users show their real dead.

Routes

Route	What
/	Landing: dig form, featured devs to haunt, their most famous dead repos
/:username	The graveyard: rank, stats, share bar, tombstone grid, "roast a friend"
/:username/:repo	One grave, with its own OG card, for "RIP " posts
/api/og/:username	1200×630 share card (?format=portrait → 1080×1350 for image posts)
/api/og/:username/:repo	1200×630 single-grave card
POST /api/dig	Counts a dig (real GitHub users only) in Firestore

Run locally

npm install
cp .env.example .env   # fill in what you need; everything is optional locally
GITHUB_TOKENS=$(gh auth token) npm run dev

Analytics and the dig counter

• Firebase Analytics tracks page views, digs, shares (by method, and whether people roast themselves or a friend), featured-dev clicks and stars. It loads after first paint and only sends events from production builds; in dev they're logged to the console. Add tracking to any element with data-track="event" plus data-track-* params.
• Dig counter: each dig increments digs/{login} in Firestore (count + last dug). Nothing about the visitor is stored. Writes go through POST /api/dig with the Admin SDK; without FIREBASE_SERVICE_ACCOUNT it's a no-op.

Deploy (Vercel)

1. Push to GitHub and import into Vercel.
2. Set GITHUB_TOKENS: one or more no-scope classic tokens, comma separated. Each adds 5k req/hour, and a cold graveyard costs 2–4 requests.
3. Set NEXT_PUBLIC_SITE_URL once you have a domain. It's printed on every share card, so pick something short.
4. Set the Firebase web config (API_KEYS, AUTH_DOMAIN, PROJECT_ID, STORAGE_BUCKET, MESSAGING_SENDER_ID, APP_ID, MEASUREMENT_ID) for Analytics, and FIREBASE_SERVICE_ACCOUNT (the service account JSON, on one line) for the dig counter.
5. Check a deployed graveyard URL in a link-preview checker, then post it in a private X draft to confirm the large card renders.