Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Execute CodeRabbit incident response procedures when reviews stop working or block PRs. Use when CodeRabbit is down, reviews are not posting, PRs are blocked by stale checks, or CodeRabbit is producing incorrect reviews. Trigger with phrases like "coderabbit incident", "coderabbit outage", "coderabbit down", "coderabbit broken", "coderabbit emergency", "coderabbit not reviewing".
.claude/skills/jeremylongshore-coderabbit-incident-runbook/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 53% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 87% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 71% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 75% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 47% | 0% |
Contain impact while preserving independent security and CI. Separate provider health, service health, config, permissions, limits, and harmful output.
references/official-docs.md and re-check any time-sensitive contract before execution.Treat Git-provider sessions, CodeRabbit web sessions, CLI credentials, and CodeRabbit API keys as separate credentials. Use only an already-approved session or secret-manager reference, never print a secret, and do not place credentials in .coderabbit.yaml, source files, logs, or deliverables.
Only documented repository authority may relax a CodeRabbit-related gate; unrelated controls remain required. Keep analysis and drafts local until approval is explicit, and record who approved the action and its scope.
A timeline, impact, containment, continuity control, recovery proof, and follow-up actions. Include source dates, unknowns, and the exact boundary between observed fact and recommendation.
| Condition | Response | |---|---| | Current contract is unclear or docs disagree | Stop mutation, cite both sources, and request owner resolution. | | Required access or approval is missing | Produce a draft and evidence plan only. | | Validation or pilot behavior differs from expectation | Restore the prior state and retain the failed evidence. | | Output contains secrets or private code | Stop, quarantine the artifact, redact it, and notify the data owner. |
Use two named human reviewers during a confirmed outage.
Stop a faulty global override and prove restored config.
references/official-docs.md.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 14,948 | 9,123 | -39% | 1 | 1 | 0% | 2,537 | 3,877 | +53% | 0 | 0 | — |
case-02 | fail→pass | 15,242 | 12,842 | -16% | 1 | 1 | 0% | 2,380 | 4,459 | +87% | 0 | 0 | — |
case-03 | fail→pass | 14,291 | 14,542 | +2% | 1 | 1 | 0% | 2,801 | 4,792 | +71% | 0 | 0 | — |
case-04 | pass→pass | 4,894 | 2,506 | -49% | 1 | 1 | 0% | 846 | 2,597 | +207% | 0 | 0 | — |
case-05 | fail→pass | 8,594 | 2,537 | -70% | 1 | 1 | 0% | 1,435 | 2,514 | +75% | 0 | 0 | — |
case-06 | fail→pass | 10,773 | 2,783 | -74% | 1 | 1 | 0% | 1,724 | 2,539 | +47% | 0 | 0 | — |
case-07 | pass→pass | 4,556 | 4,924 | +8% | 1 | 1 | 0% | 828 | 2,631 | +218% | 0 | 0 | — |
case-08 | fail→pass | 8,025 | 1,863 | -77% | 1 | 1 | 0% | 1,211 | 2,475 | +104% | 0 | 0 | — |
case-09 | fail→pass | 6,375 | 5,486 | -14% | 1 | 1 | 0% | 985 | 3,180 | +223% | 0 | 0 | — |
case-10 | fail→pass | 9,832 | 4,357 | -56% | 1 | 1 | 0% | 1,655 | 2,890 | +75% | 0 | 0 | — |
case-11 | pass→pass | 5,042 | 5,032 | -0% | 1 | 1 | 0% | 895 | 2,686 | +200% | 0 | 0 | — |
case-12 | fail→fail | 11,133 | 4,345 | -61% | 1 | 1 | 0% | 1,847 | 2,961 | +60% | 0 | 0 | — |
case-13 | fail→pass | 9,372 | 2,228 | -76% | 1 | 1 | 0% | 1,610 | 2,601 | +62% | 0 | 0 | — |
case-14 | pass→pass | 10,371 | 8,283 | -20% | 1 | 1 | 0% | 1,920 | 3,782 | +97% | 0 | 0 | — |
case-15 | fail→pass | 10,817 | 1,407 | -87% | 1 | 1 | 0% | 1,811 | 2,381 | +31% | 0 | 0 | — |
case-16 | pass→pass | 4,376 | 4,671 | +7% | 1 | 1 | 0% | 723 | 2,460 | +240% | 0 | 0 | — |
case-17 | pass→pass | 5,910 | 2,552 | -57% | 1 | 1 | 0% | 1,045 | 2,603 | +149% | 0 | 0 | — |
case-18 | fail→pass | 10,978 | 3,803 | -65% | 1 | 1 | 0% | 1,805 | 2,481 | +37% | 0 | 0 | — |
case-19 | pass→pass | 7,763 | 4,470 | -42% | 1 | 1 | 0% | 1,386 | 2,648 | +91% | 0 | 0 | — |
case-20 | pass→pass | 10,702 | 6,861 | -36% | 1 | 1 | 0% | 1,912 | 3,445 | +80% | 0 | 0 | — |
case-21 | pass→pass | 14,089 | 9,686 | -31% | 1 | 1 | 0% | 2,554 | 3,957 | +55% | 0 | 0 | — |
case-22 | pass→pass | 7,589 | 5,320 | -30% | 1 | 1 | 0% | 1,517 | 3,175 | +109% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +50 percentage points is the difference between those two pass rates over the 22 comparable cases.
The publisher has shipped newer versions since this run, so these numbers describe v1, not the version currently listed.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.