# lyssnarr

Self-hosted music discovery for the Pi5 — a Jellyseerr-style web UI that
finds new music for you. Search for artists/albums/tracks, get scheduled
AI-generated suggestions (Gemini) based on what you recently listened to
and what you add to your YouTube Music playlists, and add anything to
Lidarr with one click — or play it inline / open it in YouTube Music.

See [`doc/plan.md`](doc/plan.md) for the agreed plan and the
[Gitea wiki](https://gitea.brasse-pc.eu/brasse/lyssnarr/wiki) for
architecture notes.

## Prerequisites (dev box = Arch/Garuda)

```bash
sudo pacman -S nodejs npm     # node >= 20
```

## Build & run

```bash
git clone gitea-claude:brasse/lyssnarr.git
cd lyssnarr
npm run setup                 # installs backend + frontend deps
npm run build                 # -> ./build/ (frontend dist + backend)
npm run dev                   # dev servers (Vite + Fastify, hot reload)
npm test
```

## Runtime configuration

All secrets live in the Dockge stack `.env` on the Pi5 — never in the
repo. Key variables (full list in `doc/plan.md`):

| Var | What |
|-----|------|
| `GEMINI_API_KEY`      | Google AI Studio key for suggestion generation |
| `LIDARR_URL` / `LIDARR_API_KEY` | Lidarr instance to add music to |
| `SESSION_SECRET`      | cookie signing |
| `APP_PASSWORD_HASH`   | bcrypt hash for the single-user login |

## Deploy

CI (`.gitea/workflows/build.yaml`) builds the arm64 image on the Pi5
runner and pushes `localhost:5000/lyssnarr:latest`; the Dockge stack
`/srv/dockge-staks/lyssnarr/` runs it behind NPM as
`https://lyssnarr.brasse-pc.eu`.