A private media desk — discover movies, series, anime, music, and books, then push a release straight to your downloader.
ZME is a self-hosted, single-page web app that unifies media discovery and acquisition in one surface. Browse trending and popular titles pulled from TMDB (movies / series / anime), Open Library (books), and ListenBrainz (music); open a title for rich detail; then search your indexers for a release and hand it off to your download client — all without leaving the page.
It runs entirely on Cloudflare's edge: a Hono API on Workers serves the React SPA as static assets and keeps state in D1. There's no always-on server to babysit.
Self-hosted media stacks are powerful but fragmented: you discover something in one app, search for a release in another (Prowlarr), and queue it in a third (qBittorrent / Transmission / …). ZME collapses that loop into a single private desk:
- One place to discover across movies, series, anime, music, and books.
- One step from a title to a release — ZME queries your configured indexers (via Prowlarr) for you.
- One step from a release to a download — push it straight to your download client, no copy-pasting magnet links.
- A personal library and a download monitor to track what you saved and how downloads are progressing.
Everything is admin-managed and private: there's no public sign-up, and your API keys and indexer / downloader endpoints stay inside your own deployment.
| Title detail | My Library | Downloads |
|---|---|---|
 |
 |
 |
- Node.js ≥ 24 and pnpm 10
- A TMDB API key for discovery (configured inside the app)
- Optional, but that's the point: a Prowlarr instance for indexer search and a download client (qBittorrent / Transmission / Aria2 / ZPan)
git clone https://github.com/saltbo/zme.git
cd zme
pnpm installCreate a .dev.vars file with an auth signing secret (≥ 32 characters):
BETTER_AUTH_SECRET=replace-with-at-least-32-random-charactersThen start the dev server:
pnpm dev # → http://localhost:7171On first launch ZME opens an onboarding flow to create the first administrator. After signing in, connect your media sources, indexers, and downloaders from the Admin area — ZME doesn't run indexers or downloaders itself, it talks to the services you point it at.
wrangler secret put BETTER_AUTH_SECRET # set the auth secret on the Worker
pnpm db:migrate:remote # apply D1 migrations
pnpm deploy| Layer | Stack |
|---|---|
| Frontend | React 19, React Router 7, TanStack Query, Tailwind CSS 4, i18n (中文 / English) |
| Backend | Hono RPC API on Cloudflare Workers, serving the SPA as static assets |
| Data | Cloudflare D1 (SQLite) via Drizzle ORM; Better Auth for sessions |
| Integrations | TMDB, Open Library, ListenBrainz (discovery) · Prowlarr (indexers) · qBittorrent / Transmission / Aria2 / ZPan (downloaders) |
| Tooling | TypeScript, Biome, Vitest, Playwright, Wrangler, pnpm |
The server follows a clean, layered architecture (domain → use cases → adapters → HTTP) with the two halves meeting only through a shared API contract. The full layout, boundaries, and conventions live in the contributor guide — see below.
Contributions are welcome. The development workflow, architecture, gates, testing tiers, and database/codegen conventions are documented in CONTRIBUTING.md. The design system reference is in DESIGN.md.
ZME is intended for self-hosted, personal or household use only. It is a discovery and download-orchestration interface — it hosts and distributes no media itself, and only talks to the third-party services (metadata providers, indexers, and download clients) that you configure. You are solely responsible for how you use it and for complying with the laws and the terms of those services in your jurisdiction.
Running this project for commercial purposes is entirely at your own risk; you assume all resulting legal liability. The software is provided "as is", without warranty of any kind, as set out in the license below.