Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Apply production-ready CodeRabbit automation patterns using GitHub API and PR comments. Use when building automation around CodeRabbit reviews, processing review feedback programmatically, or integrating CodeRabbit into custom workflows. Trigger with phrases like "coderabbit automation", "coderabbit API patterns", "automate coderabbit", "coderabbit github api", "process coderabbit reviews".
.claude/skills/jeremylongshore-coderabbit-sdk-patterns/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 69% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 144% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 50% | 0% |
CodeRabbit documents REST APIs for organizations, metrics, learnings, users, roles, logs, and security. Implement only referenced operations.
references/official-docs.md and re-check any time-sensitive contract before execution.x-coderabbitai-api-key header.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.
Require Admin and security approval for keys; require separate approval for writes, roles, seats, or exports. Keep analysis and drafts local until approval is explicit, and record who approved the action and its scope.
A contract record, adapter, redacted fixtures, pagination handling, live-read receipt, and rotation owner. 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. |
Fetch paginated metrics with cursor provenance.
Reconcile partial success before bulk role changes.
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 | 12,846 | 8,153 | -37% | 1 | 1 | 0% | 2,622 | 3,343 | +27% | 0 | 0 | — |
case-02 | fail→pass | 11,956 | 8,824 | -26% | 1 | 1 | 0% | 2,398 | 3,421 | +43% | 0 | 0 | — |
case-03 | fail→fail | 15,861 | 11,601 | -27% | 1 | 1 | 0% | 3,174 | 3,820 | +20% | 0 | 0 | — |
case-04 | pass→pass | 7,343 | 6,466 | -12% | 1 | 1 | 0% | 1,412 | 2,901 | +105% | 0 | 0 | — |
case-05 | pass→pass | 8,120 | 6,228 | -23% | 1 | 1 | 0% | 1,417 | 2,687 | +90% | 0 | 0 | — |
case-06 | pass→fail | 12,124 | 9,159 | -24% | 1 | 1 | 0% | 2,227 | 3,354 | +51% | 0 | 0 | — |
case-07 | pass→pass | 7,943 | 5,618 | -29% | 1 | 1 | 0% | 1,603 | 2,757 | +72% | 0 | 0 | — |
case-08 | pass→pass | 8,687 | 2,189 | -75% | 1 | 1 | 0% | 1,635 | 2,018 | +23% | 0 | 0 | — |
case-09 | pass→pass | 11,613 | 7,277 | -37% | 1 | 1 | 0% | 2,343 | 3,093 | +32% | 0 | 0 | — |
case-10 | pass→pass | 7,543 | 4,426 | -41% | 1 | 1 | 0% | 1,254 | 2,036 | +62% | 0 | 0 | — |
case-11 | fail→pass | 7,196 | 3,384 | -53% | 1 | 1 | 0% | 1,215 | 2,050 | +69% | 0 | 0 | — |
case-12 | fail→pass | 4,910 | 3,659 | -25% | 1 | 1 | 0% | 790 | 1,929 | +144% | 0 | 0 | — |
case-13 | pass→pass | 6,943 | 4,187 | -40% | 1 | 1 | 0% | 1,233 | 2,504 | +103% | 0 | 0 | — |
case-14 | pass→pass | 6,511 | 3,370 | -48% | 1 | 1 | 0% | 1,271 | 2,314 | +82% | 0 | 0 | — |
case-15 | pass→pass | 4,080 | 1,142 | -72% | 1 | 1 | 0% | 713 | 1,799 | +152% | 0 | 0 | — |
case-16 | pass→pass | 2,330 | 1,767 | -24% | 1 | 1 | 0% | 387 | 1,958 | +406% | 0 | 0 | — |
case-17 | pass→pass | 3,133 | 1,553 | -50% | 1 | 1 | 0% | 503 | 1,875 | +273% | 0 | 0 | — |
case-18 | fail→pass | 10,974 | 6,388 | -42% | 1 | 1 | 0% | 1,879 | 2,812 | +50% | 0 | 0 | — |
case-19 | fail→pass | 9,245 | 10,871 | +18% | 1 | 1 | 0% | 1,681 | 2,769 | +65% | 0 | 0 | — |
case-20 | pass→pass | 3,472 | 1,967 | -43% | 1 | 1 | 0% | 581 | 2,011 | +246% | 0 | 0 | — |
case-21 | pass→pass | 5,947 | 2,854 | -52% | 1 | 1 | 0% | 1,137 | 2,175 | +91% | 0 | 0 | — |
case-22 | pass→pass | 10,034 | 6,265 | -38% | 1 | 1 | 0% | 2,007 | 2,860 | +43% | 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 +23 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.