Docker

Docker is the easiest way to run Streamline on a home server. The image is published for both x86 and ARM (including Raspberry Pi and Apple silicon) at ghcr.io/abhichandra21/streamline.

Set it up

git clone https://github.com/abhichandra21/streamline.git
cd streamline
mkdir -p data recommender/cache logs
cp config.local.example.yaml config.local.yaml

Put your keys in .env: TMDB_API_KEY plus one LLM key, such as ANTHROPIC_API_KEY.
Copy your export zips into data/ and set their paths in config.local.yaml, for example data/netflix/export.zip. See Watch history.

Run setup once:

docker compose run --rm streamline ./recommend setup

Start the web UI:
Terminal window
```
docker compose up -d
```
Open http://localhost:5051.

Everyday commands

Every ./recommend command works inside the container:

docker compose run --rm streamline ./recommend setup --refresh-profile
docker compose run --rm streamline ./recommend setup --refresh-data
docker compose exec streamline ./recommend "gritty British crime drama"

Update to the latest version with:

git pull
docker compose pull
docker compose up -d

If git pull says your local changes to config.yaml would be overwritten, see Settings.

Where your data lives

The container keeps nothing of its own. Everything is in the checkout folder, in the same places as a non-Docker install:

Folder or file	What it holds
`data/`	Your exports and `streamline.db` (history, ratings, watchlist, follows, searches)
`recommender/cache/`	TMDB data, descriptions, the taste profile, IMDb ratings
`logs/`	The app log
`config.yaml`	The shipped default settings; updated by `git pull`
`config.local.yaml`	Your export paths, and anything you change on the Settings page

Back up data/streamline.db; Plex plays and your ratings exist only there.

Notes

Create config.local.yaml before the first docker compose command. If it is missing, Docker creates a folder with that name instead, and the app can’t start.
The container runs as user ID 1000. On Linux, if your user has a different ID, run sudo chown -R 1000:1000 data recommender/cache logs.
Set STREAMLINE_PASSWORD in .env if anyone else can reach the server. See Running as a service.
To build the image yourself instead of pulling it, run docker compose build.