These were pre-existing failures (stale tests, not source bugs):
- getNextStep walks steps ordered by `seq`, not array order (ai-act seq 350
sits before import 400). The tests assumed array order; derive the
expectations from the seq-sorted sequence instead.
- buildDocumentScope: a document required only by the level matrix is
`mandatory` but may be `medium` priority — only trigger-mandated docs (and
the high-priority doc types) are forced to high. The test wrongly asserted
ALL mandatory docs are high; now it checks the trigger-mandated ones.
Full vitest suite: 414/414 green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Benjamin Admin
19786c96f8