Implements SC2/SC3/SC5b/SC7 + additive-field invariant for the v4.2
annotation pipeline:
Fixtures (tests/fixtures/annotation/):
- annotation-brief.md — brief-validator-clean fixture
- annotation-plan.md — plan-validator-clean (2 steps)
- annotation-review.md — review-validator-clean
- annotation-plan-large.md — 51 steps (SC3 scale fixture)
Integration tests:
- tests/integration/annotation-roundtrip.test.mjs — 7 cases:
SC2 byte-identical empty round-trip across brief/plan/review,
SC3 scale (51 steps + 100 anchors) round-trip,
SC7 parseAnchors(stripAnchors(addAnchors(...))) === [] per target.
- tests/integration/schema-rollback.test.mjs — 4 cases:
SC5b validator-FAIL -> revisionGuard rolls back byte-identical
(sha256 invariant) for brief/plan/review + cross-target sweep.
.local.bak deleted on rollback path (validator-PASS path tested
in lib/util/revision-guard tests).
- tests/lib/source-annotations.test.mjs — 6 cases mirroring
tests/lib/source-findings.test.mjs additive-field pattern: each
validator (brief/plan/review) accepts source_annotations as
additive-optional, parser extracts as array of dicts, entries
conform to documented shape, baseline forward-compat (artifacts
without source_annotations still validate).
Verify: node --test tests/integration/annotation-roundtrip.test.mjs
tests/integration/schema-rollback.test.mjs
tests/lib/source-annotations.test.mjs -> 17 pass / 0 fail.
Full npm test: 577 pass / 0 fail / 2 skipped (Docker).
Refs plan.md Step 7 + plan-critic M4 + plan-critic B4.
857 B
857 B
| type | review_version | task | slug | project_dir | brief_path | scope_sha_end | reviewed_files_count | findings |
|---|---|---|---|---|---|---|---|---|
| trekreview | 1.0 | Demo review for annotation round-trip | annotation-review-demo | .claude/projects/2026-05-09-annotation-demo | .claude/projects/2026-05-09-annotation-demo/brief.md | 0000000000000000000000000000000000000000 | 0 |
Demo review for annotation round-trip
This fixture is used by tests/integration/annotation-roundtrip.test.mjs
to verify SC2 (byte-identical empty-anchor round-trip) and SC7 (per-target
isolation against validateReview).
Executive Summary
Verdict: ALLOW. No findings. This is a synthetic fixture used to exercise the round-trip mechanics; it does not represent a real review.
Coverage
| File | Treatment |
|---|---|
| none | no diff |
Remediation Summary
No remediation needed. ALLOW.