Lyceum

A self-hosted, Netflix-style learning platform for the video courses and ebooks you already have — streamed straight from your own cloud storage, with real progress tracking, notes that sync to your Obsidian vault, gamification (streaks, badges, daily goals), and a distraction-free focus mode.

Nothing is uploaded anywhere new. Lyceum reads your existing library on whichever storage you already use (Nextcloud, ownCloud, S3-compatible storage, Google Drive, or Dropbox) and streams it on demand — no duplication, no lock-in.

Features

• Video library — a course grid with progress bars, auto-generated section grouping for courses with chapters, search, keyboard shortcuts, playback speed control, and auto-play-next.
• Book library — PDF and EPUB reading with a real reader: continuous-scroll PDF with zoom, EPUB with font/theme (light/sepia/dark) controls and a table of contents. AZW3 support is on the roadmap (see Known limitations).
• Progress tracking — per-video and per-book, with resume-where-you-left-off.
• Notes — a notes panel beside the player/reader, autosaved and pushed as markdown files into your Obsidian vault (or any folder on your storage), using a marker-section format that never clobbers content you add by hand around it.
• Gamification — separate watch and reading streaks, daily goals, rule-based badges, and a dashboard with a 28-day activity chart.
• Focus mode — a distraction-free player layout with an optional Pomodoro-style timer.
• Multi-provider storage — WebDAV (Nextcloud/ownCloud or a plain WebDAV server), S3-compatible storage (AWS S3, MinIO, Backblaze B2), Google Drive, or Dropbox. Pick one via a single env var.

Quick start

Prerequisites: Docker and Docker Compose. That's it — everything else runs inside the container.

git clone https://github.com/RegatteVarshithReddy/lyceum.git
cd lyceum
cp .env.example .env

Edit .env:

1. Set APP_PASSWORD and SESSION_SECRET to something real.
2. Set STORAGE_PROVIDER to whichever you're using, and fill in that provider's block (see below for how to get credentials for each).
3. Set VIDEO_LIBRARY_PATH / BOOKS_LIBRARY_PATH to the folders in your storage that hold your courses and books.

docker compose up -d --build

Open http://localhost:4100, log in with APP_PASSWORD, and click Rescan library on the Library and Books pages.

⚠️ Before you expose this anywhere

Lyceum has no per-user accounts, no rate limiting on the login page, and no HTTPS of its own — it's a single shared password, designed for a personal instance sitting behind a VPN (Tailscale, WireGuard) or a reverse proxy that terminates TLS. Don't put it directly on the open internet as-is.

Storage providers

Set STORAGE_PROVIDER in .env to one of webdav, s3, googledrive, dropbox, then fill in that provider's env block.

WebDAV (Nextcloud / ownCloud / plain WebDAV)

The most common choice for self-hosted setups. In Nextcloud or ownCloud: Settings → Security → Create new app password — use that, not your account password.

WEBDAV_URL=https://your-nextcloud.example.com
WEBDAV_USERNAME=you
WEBDAV_PASSWORD=<app password>
WEBDAV_NEXTCLOUD_MODE=true   # false for a plain (non-Nextcloud/ownCloud) WebDAV server

S3-compatible (AWS S3, MinIO, Backblaze B2, …)

No OAuth — just an access key. For MinIO/B2, set S3_ENDPOINT to your instance's URL and usually S3_FORCE_PATH_STYLE=true.

S3_BUCKET=my-library
S3_REGION=us-east-1
S3_ENDPOINT=              # leave blank for real AWS S3
S3_ACCESS_KEY_ID=...
S3_SECRET_ACCESS_KEY=...
S3_FORCE_PATH_STYLE=false

Google Drive

You need to register your own OAuth app — Google doesn't allow a shared client for something like this:

1. Google Cloud Console → new project → APIs & Services → Library → enable the Google Drive API.
2. APIs & Services → OAuth consent screen → External → add yourself as a test user (fine to leave the app in "Testing" status for personal use).
3. APIs & Services → Credentials → Create OAuth client ID → type Web application → add redirect URI <your PUBLIC_URL>/api/storage/google/callback.
4. Copy the client ID/secret into .env, and set GOOGLEDRIVE_ROOT_FOLDER_ID to the ID from your library folder's Drive URL (.../folders/<this part>).
5. Start Lyceum, go to Settings, click Connect Google Drive.

Heads up: this uses the drive.readonly scope (needed to see a pre-existing library — the narrower drive.file scope can't). Google treats that as sensitive, so while your OAuth app is unverified ("Testing" status, which is fine for personal use), the refresh token expires after 7 days and you'll need to reconnect periodically. Full verification removes this but requires a Google security review — not worth it for personal use.

Dropbox

1. Dropbox App Console → Create app → Scoped access, Full Dropbox access (not "App folder" — same reasoning as Drive above, it needs to see your existing library).
2. Under Permissions, enable files.metadata.read, files.content.read, files.content.write.
3. Under Settings, add redirect URI <your PUBLIC_URL>/api/storage/dropbox/callback.
4. Copy the app key/secret into .env as DROPBOX_CLIENT_ID/DROPBOX_CLIENT_SECRET.
5. Start Lyceum, go to Settings, click Connect Dropbox.

Architecture

• Backend: Node.js (Express), better-sqlite3 for all app data (progress, notes, gamification — never your library files themselves). Single container, no separate database process.
• Frontend: React + Vite, built into the backend's static file directory — one container, one port.
• Storage abstraction: every provider implements the same small interface (listDirectory, getFileStream, getFileText, putFileText, ensureDirectory) in backend/services/storage/. Video/book streaming always forwards HTTP Range requests and pipes the response directly — nothing ever gets buffered fully in memory, regardless of provider or file size.
• Readers: react-pdf (PDF.js) for PDFs, epubjs for EPUB.

Security notes

• OAuth tokens (Google Drive/Dropbox) are stored as plaintext JSON in the SQLite database (backend/data/lyceum.db), which is gitignored/dockerignored by default and never leaves your machine. Anyone who gets a copy of that file gets standing access to whatever cloud storage account is connected — back it up with the same care as any other credential.
• Both OAuth flows use a state parameter to prevent login-CSRF.
• Redirect URIs are validated by Google/Dropbox against your own registered app, per-deployment — there's no shared-app risk since every self-hoster registers their own client ID.

Known limitations

• AZW3 (Kindle) ebook support is planned but not yet implemented — the conversion pipeline (via Calibre's ebook-convert, run server-side and cached) is designed but not built.
• Course-bundle detection (a folder containing several distinct courses rather than being one course itself) needs to be configured manually via VIDEO_BUNDLE_FOLDERS — there's no reliable way to auto-detect "this is a bundle" vs. "this is one course with chapters."
• Single shared password, no multi-user support — this is built for one person's personal library, not a shared/team deployment.

License

MIT — see LICENSE.