Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →You are a PR optimization expert specializing in creating high-quality pull requests that facilitate efficient code reviews. Generate comprehensive PR descriptions, automate review processes, and ensure PRs follow best practices for clarity, size, and reviewability.
.claude/skills/dokhacgiakhoa-comprehensive-review-pr-enhance/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-17 | ✓→✗ | ▼ Worse | -20% | 0% |
| case-18 | ✓→✗ | ▼ Worse | 0% | 0% |
| case-15 | ✓→✓ | ＝ Same ✓ | -5% | 0% |
You are a PR optimization expert specializing in creating high-quality pull requests that facilitate efficient code reviews. Generate comprehensive PR descriptions, automate review processes, and ensure PRs follow best practices for clarity, size, and reviewability.
The user needs to create or improve pull requests with detailed descriptions, proper documentation, test coverage analysis, and review facilitation. Focus on making PRs that are easy to review, well-documented, and include all necessary context.
$ARGUMENTS
resources/implementation-playbook.md.resources/implementation-playbook.md for detailed templates and examples.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | fail→pass | 10,510 | 9,107 | -13% | 1 | 1 | 0% | 1,783 | 1,795 | +1% | 0 | 0 | — |
case-15 | pass→pass | 12,641 | 10,305 | -18% | 1 | 1 | 0% | 1,941 | 1,835 | -5% | 0 | 0 | — |
case-01 | pass→pass | 7,933 | 13,035 | +64% | 1 | 1 | 0% | 1,402 | 1,984 | +42% | 0 | 0 | — |
case-02 | pass→pass | 13,250 | 12,560 | -5% | 1 | 1 | 0% | 2,482 | 2,558 | +3% | 0 | 0 | — |
case-03 | fail→pass | 13,726 | 16,876 | +23% | 1 | 1 | 0% | 2,138 | 2,917 | +36% | 0 | 0 | — |
case-05 | pass→pass | 11,933 | 10,994 | -8% | 1 | 1 | 0% | 1,801 | 1,903 | +6% | 0 | 0 | — |
case-06 | pass→pass | 12,248 | 10,654 | -13% | 1 | 1 | 0% | 1,976 | 2,044 | +3% | 0 | 0 | — |
case-07 | pass→pass | 5,861 | 5,441 | -7% | 1 | 1 | 0% | 844 | 1,124 | +33% | 0 | 0 | — |
case-08 | pass→pass | 12,551 | 13,187 | +5% | 1 | 1 | 0% | 2,013 | 2,186 | +9% | 0 | 0 | — |
case-09 | pass→pass | 9,403 | 6,346 | -33% | 1 | 1 | 0% | 1,632 | 1,413 | -13% | 0 | 0 | — |
case-10 | pass→pass | 11,678 | 9,499 | -19% | 1 | 1 | 0% | 1,558 | 1,852 | +19% | 0 | 0 | — |
case-11 | pass→pass | 15,027 | 15,084 | +0% | 1 | 1 | 0% | 2,528 | 2,849 | +13% | 0 | 0 | — |
case-12 | pass→pass | 8,139 | 9,017 | +11% | 1 | 1 | 0% | 1,351 | 1,547 | +15% | 0 | 0 | — |
case-13 | pass→pass | 11,267 | 13,339 | +18% | 1 | 1 | 0% | 1,623 | 2,241 | +38% | 0 | 0 | — |
case-14 | pass→pass | 10,358 | 10,990 | +6% | 1 | 1 | 0% | 1,637 | 1,708 | +4% | 0 | 0 | — |
case-16 | pass→pass | 12,977 | 12,277 | -5% | 1 | 1 | 0% | 1,933 | 1,841 | -5% | 0 | 0 | — |
case-17 | pass→fail | 14,853 | 8,750 | -41% | 1 | 1 | 0% | 2,243 | 1,791 | -20% | 0 | 0 | — |
case-18 | pass→fail | 11,268 | 9,972 | -12% | 1 | 1 | 0% | 1,825 | 1,829 | +0% | 0 | 0 | — |
case-19 | pass→pass | 15,072 | 12,108 | -20% | 1 | 1 | 0% | 2,286 | 2,155 | -6% | 0 | 0 | — |
case-20 | pass→pass | 2,766 | 7,169 | +159% | 1 | 1 | 0% | 454 | 1,269 | +180% | 0 | 0 | — |
case-21 | pass→pass | 16,406 | 15,101 | -8% | 1 | 1 | 0% | 2,645 | 2,786 | +5% | 0 | 0 | — |
case-22 | pass→pass | 10,069 | 8,866 | -12% | 1 | 1 | 0% | 1,610 | 1,515 | -6% | 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 0 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are 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.