feat(review): guard-review waive — decoy-minting waives with rationale capture - #302
Conversation
…e capture
Capture point 2 of the corpus-growth charter (docs/benchmarks/corpus-growth.md):
guard-review waive <reviewer>[:<lens>] <fingerprint> "<rationale>" records a
persistent waive into .devkit/correctness-overrides.json with a durable audit
trail ({reviewer, lens, itemId, rationale, author, at, by:'cli'}); --list shows
active waives. Rationale is required (>=15 chars, placeholders refused) — a
waive without a reason is a bypass, not a label.
- The fingerprint printed in FAIL output IS the item id — zero gate-side
consumption changes; reconcile()/applyOverrideValve pick the entry up on the
next run. Cascade reviewers are refused (the valve only governs the pinned
single-pass reviewers; loosening that is a separate decision).
- Telemetry: a fired waive's lens item now carries disposition 'waived' AND the
rationale (capped 500 chars in the event; full text stays in the file) — the
decoy-minting signal the benchmark loop consumes.
- Concurrency (all reviewer-caught across three gate rounds): every writer of
the store — reconcile()'s env write-through AND the waive CLI — now runs
inside a shared withOverridesLock (cross-process wx lockfile; in-process
overlap impossible since both RMWs are synchronous); the stale-lock steal is
atomic via rename-to-unique; env-channel writes merge onto existing entries
instead of clobbering CLI audit fields.
- FAIL-output hint names the command beside the OVERRIDE_ env line.
- Lives in gate-engine/review/valve/ (root folder was at the 12-file fan-out
cap; overrides.mts joins valve/ when run-review.mts next changes for
substantive reasons).
465 tests green; tsc/biome clean.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
Warning Review limit reached
Next review available in: 51 minutes Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (8)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
…clone-gate ruling (#310) benchmarks-grow-from-telemetry gains its convergence record before the release that ships it: capture loop closed end-to-end (#295/#302/#303/#309, first 8 pure-telemetry rows, corpus 128), label-trust precondition met (#304: κ 0.735 post-triage, 4.2% noise floor; cleanlab floor still pending bench pred_probs), and the Target's c-CRAB/CR-Bench known-answer path recorded as falsified (#307) with the replacement candidates awaiting ratification. New axis clone-gate-non-import-code ([VALIDATED]): clones are measured over non-import code, excluded at the jscpd tokenizer — with the six-hole failure of post-hoc fragment classification recorded as the rejected road so a future simplifier can't silently re-vacuous the gate (#305/#308). Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
1 participant
Capture point 2 of the corpus-growth charter (
docs/benchmarks/corpus-growth.md) — the loop's decoy mint, fully CodeRabbit-independent.guard-review waive <reviewer>[:<lens>] <fingerprint> "<rationale>"records a persistent, audited waive;--listshows active ones. The fingerprint printed in FAIL output is the item id, so the gate needed zero consumption changes —reconcile()picks the entry up on the next run. Rationale is required (≥15 chars, placeholders refused): a waive without a reason is a bypass, not a label.Telemetry: a fired waive's lens item carries
disposition: 'waived'+ the rationale (capped 500 chars in the event, full text in the store) — that pair is the decoy-minting signal the benchmark loop consumes.Concurrency hardening — every finding below was caught by the commit gate's own correctness reviewer across three rounds, none waived:
withOverridesLock.reconcile()'s env write-through also raced the CLI → the lock moved intooverrides.mtsand now guards every writer of the store; env writes merge onto existing entries instead of clobbering CLI audit fields.Scope guard: cascade reviewers are refused (the valve only governs the model-pinned single-pass reviewers; loosening that is a separate decision). Lives in
gate-engine/review/valve/(review root was at the 12-file fan-out cap;overrides.mtsjoins it whenrun-review.mtsnext changes substantively — moving it now would drift the published implementation hash over an import line).465 tests green; tsc/biome clean.
🤖 Generated with Claude Code