Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when asked to review a PR, run a code review loop, or invoke the ralph reviewer against a pull request number or GitHub URL
.claude/skills/mikeyobrien-review-pr/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | 55% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -57% | 0% |
Run the ralph.reviewer.yml orchestration loop against a pull request. The loop checks out the PR in an isolated worktree, runs tests, reviews the diff, and produces a structured report.
/review-pr <PR number or URL>Accepts: 207, #207, or https://github.com/.../pull/207
Extract the PR number from the argument. Strip # prefix or extract from URL path.
bashralph run -H ralph.reviewer.yml -p "Review PR #<N>"
Bash tool settings:
timeout: 600000 (10 minutes)run_in_background: trueUse TaskOutput with block: true to wait for completion.
Read and print .ralph/REVIEW-REPORT.md to the conversation.
If the report file doesn't exist (loop failed before the synthesizer hat), check for and display whatever intermediate files exist:
.ralph/review-scope.md — what was scoped.ralph/review-verification.md — test results.ralph/review-findings.md — review findingsCheck that the review worktree was removed:
bashls -d .worktrees/review-<N> 2>/dev/null
Also note the presence of intermediate files (.ralph/review-scope.md, etc.) — they're useful for debugging but the user may want to clean them up later.
| Situation | Action | |-----------|--------| | Ralph exits non-zero | Display error output. Suggest re-running with RALPH_DIAGNOSTICS=1 | | Report file missing | Display intermediate files that do exist (scope, verification, findings) | | PR argument missing | Ask the user for the PR number | | PR argument unparseable | Ask the user to provide a bare number, #N, or full GitHub URL |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 3,577 | 2,536 | -29% | 1 | 1 | 0% | 607 | 897 | +48% | 0 | 0 | — |
case-02 | fail→fail | 8,726 | 5,104 | -42% | 1 | 1 | 0% | 1,444 | 854 | -41% | 0 | 0 | — |
case-03 | fail→fail | 7,490 | 4,133 | -45% | 1 | 1 | 0% | 1,357 | 833 | -39% | 0 | 0 | — |
case-04 | pass→fail | 8,093 | 7,540 | -7% | 1 | 1 | 0% | 1,558 | 1,178 | -24% | 0 | 0 | — |
case-05 | fail→fail | 2,246 | 3,069 | +37% | 1 | 1 | 0% | 336 | 757 | +125% | 0 | 0 | — |
case-06 | fail→fail | 3,180 | 5,014 | +58% | 1 | 1 | 0% | 545 | 918 | +68% | 0 | 0 | — |
case-07 | fail→fail | 5,060 | 4,346 | -14% | 1 | 1 | 0% | 890 | 835 | -6% | 0 | 0 | — |
case-08 | fail→pass | 6,496 | 1,570 | -76% | 1 | 1 | 0% | 568 | 881 | +55% | 0 | 0 | — |
case-09 | fail→pass | 5,367 | 2,350 | -56% | 1 | 1 | 0% | 828 | 1,024 | +24% | 0 | 0 | — |
case-10 | fail→pass | 8,433 | 3,304 | -61% | 1 | 1 | 0% | 1,488 | 1,211 | -19% | 0 | 0 | — |
case-11 | fail→pass | 9,109 | 3,568 | -61% | 1 | 1 | 0% | 1,760 | 1,275 | -28% | 0 | 0 | — |
case-12 | fail→pass | 10,823 | 1,727 | -84% | 1 | 1 | 0% | 2,061 | 877 | -57% | 0 | 0 | — |
case-13 | fail→pass | 7,984 | 2,305 | -71% | 1 | 1 | 0% | 1,410 | 1,019 | -28% | 0 | 0 | — |
case-14 | fail→pass | 3,504 | 1,656 | -53% | 1 | 1 | 0% | 579 | 872 | +51% | 0 | 0 | — |
case-15 | fail→pass | 7,281 | 2,488 | -66% | 1 | 1 | 0% | 1,406 | 1,033 | -27% | 0 | 0 | — |
case-16 | pass→pass | 7,018 | 1,376 | -80% | 1 | 1 | 0% | 1,156 | 780 | -33% | 0 | 0 | — |
case-17 | pass→pass | 8,940 | 1,811 | -80% | 1 | 1 | 0% | 1,715 | 942 | -45% | 0 | 0 | — |
case-18 | fail→pass | 5,510 | 1,614 | -71% | 1 | 1 | 0% | 1,083 | 814 | -25% | 0 | 0 | — |
case-19 | fail→fail | 3,773 | 4,570 | +21% | 1 | 1 | 0% | 634 | 849 | +34% | 0 | 0 | — |
case-20 | fail→pass | 11,042 | 1,480 | -87% | 1 | 1 | 0% | 1,962 | 777 | -60% | 0 | 0 | — |
case-21 | fail→pass | 6,428 | 3,126 | -51% | 1 | 1 | 0% | 983 | 1,162 | +18% | 0 | 0 | — |
case-22 | fail→pass | 7,153 | 2,310 | -68% | 1 | 1 | 0% | 1,093 | 1,010 | -8% | 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, and 16 counted toward the lift figure. The other 6 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +50 percentage points is the difference between those two pass rates over the 16 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.