| name | review-tool-volume-management |
| description | Manage rawsql-ts external review-tool usage for CodeRabbit or similar AI review bots when PRs are stacked, small, generated, broad, rate-limited, or likely to consume review quota before the branch is ready. |
Review Tool Volume Management
Use this skill before requesting external AI review, when CodeRabbit reports rate limits, or when a branch is being updated repeatedly before it is ready for human review.
Workflow
- Decide whether the PR is ready for external review:
- implementation complete,
- focused verification run,
- PR readiness body validated,
- self-review blockers resolved or surfaced.
- If the PR is still changing quickly, avoid manual review triggers until the next ready checkpoint.
- For stacked or closely related small PRs, request review on the PR where the source decision is clearest, then use the review result to update the rest.
- For generated or broad diffs, use
broad-generated-diff-review-packet before requesting bot review.
- When a rate-limit comment appears, record it as review coverage unavailable, not as a code finding.
- If review is still needed after the refill window, trigger one review after the branch and PR body are stable.
- Do not treat
COMMENTED bot review states as approval.
PR Notes
Mention review-tool limits only when they affect review confidence. Keep the note short:
External AI review was rate-limited; repository checks and self-review are the current evidence.
External AI review requested after readiness validation.
Constraints
- Do not spend review quota on WIP pushes when local checks or PR readiness are still failing.
- Do not hide skipped or rate-limited review coverage if the final claim depends on it.
- Do not copy bot marketing, tips, or generated poems into repository evidence.