Vodloader

Captures Twitch streams (upload to YouTube not implemented yet). Runs as a Docker stack: the vodloader app plus nginx as a reverse proxy.

Quick start

1. Clone the repo.

2. Copy example.env to .env and set your values (Twitch credentials, WEBHOOK_HOST, API_KEY, etc.). See app/README.md for all environment variables.

3. Optionally edit compose.yml to change the videos volume (default: /mnt/media/Capture/vodloader).

4. Run:

   docker compose up -d

The API and web UI are served on port 8000 (nginx). See app/README.md for API endpoint documentation. Database and captured videos are stored in ./database and the mounted videos volume respectively.

NVIDIA GPU

The stack is configured to pass all NVIDIA GPUs into the vodloader container (deploy.resources.reservations.devices). Install NVIDIA Container Toolkit on the host so Docker can use the GPU. The vodloader service also uses ipc: host and init: true (recommended for Playwright Chromium). The app currently uses software encoding (libx264); to use hardware encoding (e.g. h264_nvenc) you would need to configure the transcoding pipeline accordingly.

Repository layout

Path	Description
app/	Main Python application. See app/README.md for setup, config, and development.
nginx/	Nginx reverse proxy; forwards traffic to the app.
compose.yml	Docker Compose stack (vodloader + nginx).
example.env	Example environment file; copy to .env and configure.

License

See LICENSE.