Skip to content

ADR-0011: Product definition: parity with Huntarr and newtarr

  • Status: accepted
  • Date: 2026-10-17

Context

No document said what a complete SnatchArr is from the point of view of a Huntarr or newtarr user. Issue #321 compared the three as a parity matrix:

  • Huntarr v9.4.2 (2026-02-23), the last release before its repository was removed, read from a public archive mirror.
  • newtarr v1.0.0 (2026-02-24), the elfhosted fork of Huntarr v6.6.3.
  • SnatchArr as of main e5e624fd.

The matrix found 17 expectations already covered and 8 open gaps. Each open gap was decided as in or out of scope.

Decision

SnatchArr is complete for a Huntarr or newtarr user when it covers the expectations below.

Covered

Covered already, as listed in #321:

  • Apps: Sonarr, Radarr, Lidarr, Readarr (frozen, ADR-0008), and Whisparr v2 and v3, with several instances per app.
  • Searching:
  • missing and cutoff-unmet searches, with items per cycle and a cycle interval
  • an hourly cap per instance
  • queue-aware pausing through download client backpressure (ADR-0006)
  • schedules and pause windows
  • Sonarr season packs
  • Memory and history: processed-item memory with reset, history, and live activity over SSE.
  • Links: a read-only Prowlarr link.
  • Access: login, reverse-proxy and OIDC sign-in (Single Sign-On), and a REST API with API keys described in OpenAPI 3.1.
  • Notifications: Discord, Slack, Telegram, ntfy, Gotify, Teams and signed webhooks.
  • Installs: compose, a Helm chart and ArgoCD.
  • Backups: delegated to CloudNativePG or the compose volume (Deployment).

In scope, open

Expectation Issue
Pushover, Pushbullet, email (SMTP), Notifiarr and others through one Apprise channel #348
In-app log viewer for api and worker logs #349
Radarr minimum availability filter: skip movies Radarr reports as not yet available #350

Out of scope

Expectation Reason
TOTP two-factor login OIDC sign-in hands two-factor login to the identity provider (Single Sign-On). A second-factor store in SnatchArr would duplicate it and add secrets to keep.
Recovery key or password reset OIDC and proxy sign-in map to the single admin account, so a locked-out admin signs in through them. A recovery key would be one more long-lived credential to store and leak.
Local-network login bypass It trades authentication for a network address that a reverse proxy, VPN or container network can make ambiguous. Trusted proxy-header sign-in covers the convenience case with an explicit trust boundary.
Plex sign-in Plex sign-in goes through Plex's own account API, not OIDC, so it would need a Plex-specific auth path beside OIDC and proxy sign-in. Users who want Plex identities put an identity provider in front.
Unraid template, native Windows or macOS installs The signed and attested container images are the release unit, and an Unraid host runs containers. Native builds would add platforms without that provenance and outside the tested matrix.

Out of scope by design, as already recorded:

  • Swaparr (stalled downloads): ADR-0004. Cleanuparr owns it (ADR-0007). (Note: ADR-0012 supersedes the Cleanuparr link; see #362.)
  • Movie Hunt, TV Hunt, Index Master, NZB Hunt, Requestarr: these replace Radarr, Sonarr, Prowlarr, SABnzbd and Seerr. SnatchArr drives the *arr apps and links Seerr and Prowlarr; it does not replace them.

Consequences

  • Positive: the scope is written down. A feature request is checked against this record, and a parity gap either has an issue above or a stated reason.
  • Negative: users migrating from Huntarr lose TOTP, Plex sign-in, LAN bypass and native installs. They have to use an identity provider or a container host instead.
  • Revisit when: a decided-out row gets a request with a use case that its reason does not answer, or a new Huntarr-family release adds an expectation.