Run modes and pipeline

Full reconciliation

A full run scans the complete selected source scope, maps all eligible items, reads the corresponding target state, plans enabled features, and reconciles managed custom lists. The first successful live full run establishes the profile baseline. A dry-run does not establish it.

Use full reconciliation for the first sync, after resetting profile data, after substantial scope or mapping changes, and when remote target drift must be repaired.

Incremental delta

Incremental runs require a live full baseline. They compare feature-specific source fingerprints, restore cached mappings, and avoid target reads for unchanged features. A relevant source change still reads current target state before planning so remote manual edits and protection rules are respected.

Incremental is an API-saving optimization, not a guarantee that arbitrary remote changes are discovered. Use full reconciliation to audit target-only drift.

Webhook runs

Plex events enter the same targeted incremental pipeline. Events are filtered to enabled profiles and selected libraries, deduplicated, and coalesced for five seconds so related playback/progress/rating events can share one run. See Webhooks.

Generic pipeline

1. Load the profile and resolved feature schema.
2. Validate source/target settings and credentials; refresh supported tokens.
3. Scan or restore source items and optional feature enrichment.
4. Apply source-filter features and media compatibility filtering.
5. Map items, including cached or one-to-many AniBridge segments.
6. Read current target state in provider-sized batches.
7. Plan deterministic normalized operations and resolve conflicts.
8. Persist a dry-run or write live operations with target pacing and retries.
9. Confirm only successful feature state and fingerprints.
10. Reconcile managed custom lists after normal item operations.

Long-import safety analysis

Before a live full run, Brokerr can perform a fast source-only analysis. It estimates potential diary, list, watchlist, rating, and other writes from source counters without loading per-film Plex cloud history or target state. The UI uses target pacing to show an approximate minimum duration.

The confirmed execution creates an exact persisted plan. If its write phase is interrupted, Brokerr can resume remaining operations without rescanning the complete Plex cloud history. Configuration changes make the estimate or plan stale.