SeaDex Mirror

An unofficial static mirror of SeaDex (source), the anime release recommendation index. Built as a static site so the data stays accessible even when the original is down.

Live site: seadex.pages.dev

How it works

flowchart LR
    A[SeaDex API] --> B[Build Script]
    C[AniList API] --> B
    B --> D[Static JSON]
    D --> E[Vite Build]
    E --> F[Cloudflare Pages]

Loading

The build script pulls all public entry and torrent data from the SeaDex PocketBase API, enriches it with AniList metadata (titles, covers, scores, relations), and writes everything as static JSON files. Vite bundles the frontend, and the result gets deployed to Cloudflare Pages as a fully static site.

No database, no server-side runtime, no API keys required for browsing.

Getting started

Prerequisites: Node.js 24+

git clone https://github.com/EithonX/seadex-mirror.git
cd seadex-mirror
npm install
npm run dev

To build with live data from upstream:

npm run data:build
npm run dev

Commands

Command	What it does
npm run dev	Start the Vite dev server
npm run build	Build a deployable site snapshot
npm run build:frontend	Build only the frontend
npm run data:build	Fetch live data from SeaDex + AniList
npm run build:site	Build data and frontend together
npm run deploy	Deploy to Cloudflare Pages
npm run typecheck	Run TypeScript type checking
npm run verify:mirror-data	Verify generated JSON structure
npm run verify:frontend-build	Verify production build output

Data pipeline

The build script (scripts/build-static-data.mjs):

1. Fetches entry IDs from the listIDs endpoint
2. Fetches all entries with expanded torrent data (expand=trs)
3. Verifies parity between the two data sources
4. Resolves relative torrent URLs against https://releases.moe
5. Fetches AniList metadata for each entry
6. Enriches franchise relation data for entry detail pages
7. Writes static JSON files into frontend/public/mirror-data/

Output files

File	Used by
catalog.json	Home page — table filtering and search
entries/<alId>.json	Individual entry pages
sheet-workbook.json	Sheet view (lazy-loaded separately)
status.json	Mirror freshness indicator

Unchanged-upstream behavior

The data builder supports two modes when upstream data hasn't changed:

• --onUnchanged=skip — For scheduled CI rebuilds. Exits early without rebuilding or deploying.
• --onUnchanged=materialize — For deploy builds. Reconstructs local data from cache when upstream is unchanged but local files are missing (e.g. fresh CI checkout).

Deployment

Deployed to Cloudflare Pages through GitHub Actions using Wrangler Direct Upload.

Required secrets

Secret	Purpose
CLOUDFLARE_ACCOUNT_ID	Cloudflare account identifier
CLOUDFLARE_API_TOKEN	API token with Pages deploy permissions

Optional

Secret / Variable	Purpose
CLOUDFLARE_PAGES_PROJECT_NAME	Pages project name (default: seadex)
ANILIST_ACCESS_TOKEN	Authenticated AniList access for higher rate limits

Workflows

rebuild-mirror.yml runs every 12 hours. Only rebuilds and deploys when upstream data actually changed. Supports a manual force trigger.

deploy-site.yml runs on pushes to main that touch frontend code, scripts, or build config. Uses --onUnchanged=materialize so frontend-only changes still deploy with complete data.

Cache policy

Path	TTL
catalog.json, status.json, sheet-workbook.json	60 seconds
entries/*	15 minutes
Hashed frontend assets	Immutable

Contributing

See CONTRIBUTING.md for setup instructions and guidelines.

License

GNU General Public License v3.0

Credits

• SeaDex (source) — the upstream project this mirrors
• AniList — anime metadata