Jobs

A job is one automated discovery pipeline:

provider → discovery type → rule set → scoring → delivery target

You can run one job or combine several providers and strategies. Each job remains independently configurable and observable.

Create a job

Open Jobs and select New job.
Choose a ready recipe or open Custom Job.
Recipes create a disabled job with an explicit provider, cadence, and job delivery cap. Recipes with meaningful filtering create dedicated rules; recipes without overrides reuse the media default.
Review the automatic preview, adjust the job or its rules, then enable it.

Unavailable recipes remain visible and identify the connection they need. Recipes never silently select a different configured provider.

Built-in recipes

Outcome	Provider	Default cadence
Balanced Trending movies and shows	TMDB	Daily
New & Well Rated movies	TMDB	Daily
Anticipated With Approval	Trakt + Jellyseerr/Seerr	Daily
Documentary, Science-Fiction, Family, Reality TV, Anime	TMDB	Weekly
Current TV	TMDB	Daily
All-Time Classics	Trakt	Weekly
Personal Watchlist	Connected TMDB account	Every 2 hours

Every recipe is versioned, disabled by default, and preview-first. Creating the same recipe again makes a separately named rule set rather than changing the first job’s policy.

Discovery types

Type	Purpose
Trending	Content currently receiving attention
Popular	Broad provider popularity
Watched	Most watched for a selected period
Collected	Most collected for a selected period
Favorited	Most favorited for a selected period
Played	Most played for a selected period
Anticipated	Upcoming anticipated releases
Box office	Current movie box-office discovery
Smart Popular	Popularity with an adaptive rating threshold
List or Watchlist	A provider-owned public list or personal watchlist
Recommendations	One-hop TMDB recommendations from explicit seed IDs or a provider seed list

Support varies by provider and media type; the Jobs editor is the source of truth for valid combinations.

List jobs store a provider owner, list ID, or slug rather than an arbitrary web URL. They use the same preview, rules, scoring, duplicate checks, Activity Entries, and delivery limits as every other job. Removing an item from the source list never deletes media that Blockbusterr already delivered.

Public Trakt and TMDB lists need their provider application credentials. MDBList lists and the personal MDBList watchlist use an MDBList API key. Personal Trakt and TMDB watchlists require connecting that provider account in Settings; a public Trakt user’s watchlist can instead use the account name in the job. Use Check source before saving to verify the identifier and see a movie/show sample without delivering anything.

Letterboxd public lists and watchlists are available only after explicitly enabling experimental scraping in Settings. Its official API requires selective approval, so this fallback may be blocked or break whenever Letterboxd changes its pages. It never accesses private content and fails instead of guessing a title match. MDBList remains the reliable bridge for lists imported from Letterboxd or IMDb.

Recommendation jobs accept up to 20 explicit TMDB movie or show IDs, an existing provider list, or both. Seed lists are normalized to TMDB identities; unmapped entries are ignored and a job with no usable seeds fails clearly. Seeds and results are deduplicated, seed titles are excluded, and results are never recursively expanded.

Job settings

Source: TMDB, Simkl, Trakt, or MDBList when supported.
Media: Movie or show.
Content limit: Maximum candidates fetched before rules are evaluated.
Job delivery cap: Maximum winners a job contributes in one standalone run or ranked cycle; 0 is unlimited.
Period: Weekly, monthly, yearly, or all time for period-based types.
Rule set: The reusable policy used by preview and execution.
Sync interval: Optional Go duration or five-field cron override.
Mode: Optional direct or jellyseerr delivery override.
Availability and monitoring: Optional Radarr/Sonarr overrides.
Sonarr series type: standard, anime, or daily for show jobs.
Repeat handling: Optional override for when previously delivered and later removed media becomes eligible again.
Ranked selection: Opt a job into a shared cycle and optionally reserve a minimum number of picks.

Shared and job-specific rules

Jobs reference one media-compatible rule set. Reuse a rule set when jobs should stay synchronized, or use Make job-specific copy to clone the effective policy for one job. See Rules.

The discovery limit controls how many candidates are evaluated. The job delivery cap controls how many winners one job can contribute. Ranked-selection winner slots cap the shared cycle, while global movie and show ceilings remain the final rolling safety limit. Configure shared membership and ceilings under Settings → Automation.

Ranked selection cycles

Ranked selection is opt-in under Settings → Automation and requires content scoring. Membership can be managed there or from an individual job. Participating jobs use the shared cycle cadence instead of their own schedule. At each cycle, Blockbusterr fetches every participating job, applies its assigned rules, merges duplicate media by canonical provider ID, and selects the highest normalized scores up to the movie and show winner slots.

Optional job minimums are allocated first and are validated against cycle capacity. Remaining slots go to the highest scores with provider rank and stable IDs used as deterministic tie-breakers. Preview reports found, rule-rejected, already-present, repeat-blocked, job-capped, duplicate, below-cutoff, and globally blocked candidates. A provider failure is shown for that job while successful providers still compete. Global rolling delivery ceilings remain final safeguards, and all participating jobs retain their normal Activity Entries and Job Runs.

Preview, run, and schedule

Duration schedules run once when Blockbusterr starts and then repeat at the configured interval. Cron schedules wait for their next matching time and do not run at startup.

Preview discovers and evaluates candidates without delivering them.
Run now executes the complete pipeline immediately.
Enable/Disable controls scheduled execution without deleting the job.
Delete removes the dynamic job after confirmation.

Manual runs ignore the schedule but use the same provider, rules, scoring, and delivery configuration. Participating ranked-selection jobs cannot run independently because doing so would bypass shared competition and cycle limits.

Delivery modes

In Direct mode, accepted movies go to Radarr and shows go to Sonarr. In Jellyseerr mode, accepted titles are requested through Jellyseerr or Seerr, which applies its own approval permissions before forwarding them. See Integration Modes.

The Anime Discovery recipe uses normal show rules plus Sonarr’s anime series type. TMDB must provide a TVDB identity and Sonarr must return that exact series; Blockbusterr reports a mapping failure instead of delivering a title-only guess.

Understand a run

The Activity page has two complementary views:

Activity Entries: individual titles with outcome, score, time, source job, and decision details.
Job Runs: each execution with found, passed, added, requested, rejected, skipped, and failed counts plus the outcome distribution. Added means Blockbusterr sent the title directly to Radarr or Sonarr; Requested means it submitted the title through Jellyseerr or Seerr. Outcomes with a zero count are hidden.

Use Preview while tuning rules, Job Runs to spot pipeline-level problems, and Activity Entries to explain one title.

Upgrade legacy jobs

On first v2 startup, enabled v1 named job sections are upgraded automatically. Blockbusterr creates equivalent dynamic jobs, assigns the appropriate default rule set, preserves supported settings, saves the new configuration atomically, and disables the migrated legacy entries. The untouched v1 configuration remains available as config.yaml.backup.

Review the generated jobs and rules after startup. Do not delete the automatic backup until the installation has completed at least one expected schedule cycle. See Upgrading to v2.