Loom

Unified media automation — Radarr + Sonarr + Prowlarr in one modern, container-native, observable platform. Single Go binary, SQLite by default, Postgres-ready, K8s-native.

Status: alpha. The first tagged release (v0.1.0) is out — published binaries and container images are available. APIs and schema may still change. See the roadmap for what's done, what's in flight, and what's next. Issues and PRs welcome.

Loom is a from-scratch, like-for-like replacement of the *arr stack (Radarr, Sonarr, Prowlarr) with a modernised UI, first-class observability, and a single binary that runs anywhere. It is wire-compatible with the existing self-hosted media ecosystem, so Overseerr, Jellyseerr, Bazarr, Notifiarr, Tautulli, Plex, Jellyfin and Emby keep working unchanged.

Name rationale — "Loom": short, catchy, evokes weaving disparate threads (indexers, download clients, metadata providers, libraries) into a uniform fabric. Working name; trivial to rename later.

The following surfaces will be wire-compatible so existing clients keep working:

• /api/v3/* — Radarr v3 (Overseerr, Jellyseerr, Notifiarr).
• /api/v3/* and v4 — Sonarr (Overseerr, Bazarr).
• /api/v1/* — Prowlarr v1 (used by Sonarr/Radarr-as-clients).
• Cardigann YAML indexer definitions (Jackett / Prowlarr ecosystem) — see docs/indexers-cardigann.md.
• Newznab / Torznab (inbound and outbound).
• Download clients: qBittorrent, Transmission, Deluge, rTorrent, SABnzbd, NZBGet.

Alpha — the server boots and serves health, metrics, and the /api/v1/* endpoints (movies, series, indexers, search, downloads, settings). The end-to-end search → grab → download → import pipeline works today.

docker run --rm -p 1925:1925 \
  -v /opt/loom/config:/config \
  -v /mnt/media:/media \
  ghcr.io/ebenderooock/loom:latest

Open http://localhost:1925/healthz. Images are published to ghcr.io/ebenderooock/loom (:latest and per-release tags such as :0.1.0) on every tagged release.

Requires Go 1.26+ and Make. This path produces a working binary against the current master.

git clone https://github.com/ebenderooock/loom.git
cd loom
make build
mkdir -p ./run
LOOM_CONFIG_DIR=./run \
LOOM_DATA_DIR=./run \
LOOM_STORAGE_SQLITE_PATH=./run/loom.db \
  ./dist/loom serve
# in another terminal:
curl -s http://localhost:1925/api/v1/system/status
# {"buildDate":"...","commit":"...","engine":"sqlite","version":"..."}

git clone https://github.com/ebenderooock/loom.git
cd loom
docker compose up -d
# Loom        http://localhost:1925
# qBittorrent http://localhost:8080
# Prometheus  http://localhost:9090
# Grafana     http://localhost:3000  (admin/admin)

See docs/deployment.md for Kubernetes (Phase 11), bare-metal binaries, and reverse-proxy snippets (Traefik / Caddy / Nginx). For Synology NAS, see docs/synology.md.

The docs/ directory is the source of truth.

• Configuration reference
• Architecture
• Deployment
• Synology (DSM) install
• Observability
• Storage
• API (and api/openapi/loom.yaml)
• Indexers — aggregator (Newznab/Torznab)
• Indexers — rate limits and retry/backoff
• Download clients
• Security
• Development setup
• Contributing & code style
• Architecture decision records

The 11-phase delivery plan lives in ROADMAP.md. The core media-automation pipeline (search → grab → download → import/organize for movies and TV) is working and shipped in v0.1.0; wire-compatibility APIs, request portal, and the migration importer are in progress.

PRs welcome — please read CONTRIBUTING.md and docs/contributing-style.md first.

No upstream code copied from Radarr / Sonarr / Prowlarr. Loom is a clean-room reimplementation that references public API specs, docs, and user-facing config files only.

For security issues see SECURITY.md.

AGPL-3.0. Same license family as the upstream *arr stack.

Loom is not a fork. It is a clean-room reimplementation that honours the years of community work in Radarr, Sonarr, Prowlarr, Jackett, Cardigann, TRaSH-Guides and the broader self-hosted media community. Thank you.