ERD_INDEXERS.md defines a hard-blocker migration acceptance bar, but the checklist still left Final acceptance criteria (all hard blockers) pass unchecked after the underlying API, Torznab, import, and rollback coverage had already landed across multiple slices.
We needed one explicit closeout step that ties the current evidence back to the ERD’s go/no-go criteria so the remaining unchecked items stay limited to non-hard-blocker follow-up work.
Added tests/specs/api/indexers-final-acceptance.spec.ts as a focused acceptance aggregation test.
The new spec verifies the hard-blocker user path remains:
explicit for invalid Torznab queries,
explicit for missing downloads,
explicit for missing import secrets,
reversible with no downstream app mutation surface.
The checklist is updated to mark final acceptance complete while preserving the still-open non-hard-blocker parity gaps for app sync UX, app-scoped category overrides, broader import UX, and health notifications.
No production observability changes were required.
Acceptance evidence continues to rely on existing import, Torznab, and rollback endpoint behavior plus the previously shipped health/explainability surfaces.
Risk is low because this change closes an acceptance gap with additive verification and documentation rather than altering runtime behavior.
If any acceptance assumption regresses, rollback is a straightforward revert of this ADR, the acceptance spec, and the checklist update while keeping the earlier feature slices intact.
Alternative considered: leave final acceptance unchecked until every non-hard-blocker parity item landed. Rejected because the ERD separates hard blockers from follow-up UX parity, and the repo already has the necessary migration-safety evidence to close the hard-blocker gate now.