Search films and TV, save them, and never lose your place in a series.
Progress tracked down to the episode you stopped on, streaming availability in your country, and suggestions drawn from your own list.

Live demo · Report a bug · Run it locally

---

What it does

• Sign in with Google — your list is private to your account.
• Search and trending, proxied server-side so the TMDB token never reaches the browser.
• Suggestions from your list — rows of "because you watched X", built from what you actually finished rather than from what is popular this week.
• Season and episode tracking — a bookmark down to the episode, and nothing that has not aired can be ticked off, so a show is never marked watched ahead of broadcast. Finishing the aired seasons marks you caught up, and a new season brings the show back on its own. (design notes)
• Where to watch — streaming, rental and purchase options for your country.
• Coming soon — pending films and seasons grouped by how soon they arrive.
• Auto-delete — optionally clear watched titles off your list after 7, 30 or 90 days. Off by default, each card warns you for the last week with a one-tap Keep, and anything with a new season on the way is never touched.

How it works

flowchart LR
    B["🖥️ Browser"] --> W["☁️ Cloudflare Worker<br/>SvelteKit app"]
    W --> D[("🗄️ Turso")]
    W --> T["🎬 TMDB"]
    W --> G["🔑 Google OAuth"]

Loading

The browser talks to the Worker and nothing else — every credential stays inside it.

Two conventions carry most of the codebase: decision logic lives in src/lib/domain as pure functions with their own unit tests, and anything under src/lib/server can never reach the browser. Upkeep (refreshing season data, deleting expired titles) runs on page load rather than on a timer, so there is no cron job to operate.

Stack: SvelteKit 5 · TypeScript · Tailwind 4 · Drizzle + Turso (libSQL) · Cloudflare Workers

Running locally

Needs Node 22+, a free TMDB token and a Google OAuth client.

npm install
cp .env.example .env    # fill in the values below
npm run db:migrate
npm run dev             # http://localhost:5173

Variable	Description
TMDB_ACCESS_TOKEN	TMDB v4 API Read Access Token (Bearer).
DATABASE_URL	file:local.db locally, or a libsql://…turso.io URL in production.
DATABASE_AUTH_TOKEN	Turso auth token. Leave empty for the local file database.
GOOGLE_CLIENT_ID	OAuth 2.0 client ID.
GOOGLE_CLIENT_SECRET	OAuth 2.0 client secret.
OAUTH_ORIGIN	Optional. Public origin for the redirect URI; derived from the request if unset.

⚠️ .env is git-ignored and must never be committed.

Google sign-in: in the Cloud Console, create an OAuth client ID for a Web application and add one redirect URI per environment — http://localhost:5173/auth/google/callback and https://<your-domain>/auth/google/callback. Only openid, profile and email are requested, and no Google tokens are stored.

Other scripts: npm run lint, npm run check, npm run test:unit, npm run test:e2e.

npm run brand regenerates the README banner from docs/brand/banner.html. It is drawn rather than painted — same palette, typefaces and film mark as the app — so editing that file and re-running the script is how the image changes.

npm run gen regenerates worker-configuration.d.ts from the bindings in wrangler.jsonc. Run it when you add or change a binding, and commit the result. Run it on a clean tree: wrangler types adds an extra GlobalProps interface pointing at the built worker when .svelte-kit/cloudflare/_worker.js happens to exist, so regenerating after a build produces a file that differs from the one CI generates.

Deployment

Pushes to main deploy to Cloudflare automatically once CI passes. That needs one repository secret, CLOUDFLARE_API_TOKEN (Cloudflare → My Profile → API Tokens → template Edit Cloudflare Workers).

First-time setup:

turso db create nextsode   # put the URL + token in .env, then:
npm run db:migrate           # apply the schema to the remote database

# Production secrets are not read from .env:
npx wrangler secret put TMDB_ACCESS_TOKEN
npx wrangler secret put DATABASE_URL
npx wrangler secret put DATABASE_AUTH_TOKEN
npx wrangler secret put GOOGLE_CLIENT_ID
npx wrangler secret put GOOGLE_CLIENT_SECRET

To deploy by hand: npm run build && npx wrangler deploy.

Security

• Secrets live only in src/lib/server/*, which SvelteKit guarantees is never bundled for the client.
• Sign-in uses OAuth 2.0 with PKCE and a state parameter; the database stores only a SHA-256 hash of the session token, so a dump cannot be replayed as a session.
• Every watchlist read and write is scoped by the session's user id — item ids travel through the browser, so knowing one is not enough to touch someone else's list.
• Input is length- and range-checked before it reaches the database, and season targets are always resolved server-side.
• Personal responses are private, no-store; only the public TMDB proxies are edge-cached.

License

Private project — all rights reserved (for now).