Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when reviewing code changes, checking PRs, or evaluating code quality.
.claude/skills/aiskillstore-reviewing-code/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -69% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -69% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -59% | 0% |
bash# Run ALL checks at once (tests + linting) bpsai-pair ci # Validate project structure bpsai-pair validate # See what changed git diff main...HEAD --stat git diff main...HEAD
markdown## Code Review: [Description] ### Summary Brief assessment. ### Must Fix 1. **[File:Line]** - Issue and fix ### Should Fix 1. **[File:Line]** - Suggestion ### Consider 1. **[File:Line]** - Optional improvement ### Positive Notes - What was done well ### Verdict - [ ] Approve - [ ] Approve with comments - [ ] Request changes
bash# Find debug statements git diff main...HEAD | grep -E "print\(|breakpoint|pdb" # Find TODOs in changes git diff main...HEAD --name-only | xargs grep -n "TODO\|FIXME" # Check for secrets git diff main...HEAD | grep -iE "password|secret|api.?key|token"
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 8,341 | 16,477 | +98% | 1 | 1 | 0% | 566 | 558 | -1% | 0 | 0 | — |
case-02 | fail→fail | 8,638 | 14,451 | +67% | 1 | 1 | 0% | 1,481 | 589 | -60% | 0 | 0 | — |
case-03 | fail→fail | 4,879 | 6,032 | +24% | 1 | 1 | 0% | 707 | 664 | -6% | 0 | 0 | — |
case-04 | pass→fail | 15,706 | 4,615 | -71% | 1 | 1 | 0% | 1,636 | 1,056 | -35% | 0 | 0 | — |
case-05 | fail→pass | 9,632 | 6,625 | -31% | 1 | 1 | 0% | 1,670 | 516 | -69% | 0 | 0 | — |
case-06 | fail→pass | 10,696 | 1,711 | -84% | 1 | 1 | 0% | 1,704 | 530 | -69% | 0 | 0 | — |
case-07 | pass→pass | 13,670 | 4,573 | -67% | 1 | 1 | 0% | 1,503 | 1,103 | -27% | 0 | 0 | — |
case-08 | fail→pass | 9,265 | 10,016 | +8% | 1 | 1 | 0% | 1,604 | 1,100 | -31% | 0 | 0 | — |
case-09 | fail→pass | 14,790 | 14,393 | -3% | 1 | 1 | 0% | 1,758 | 1,994 | +13% | 0 | 0 | — |
case-10 | pass→pass | 12,789 | 13,409 | +5% | 1 | 1 | 0% | 1,839 | 1,657 | -10% | 0 | 0 | — |
case-11 | pass→pass | 24,273 | 11,967 | -51% | 1 | 1 | 0% | 2,059 | 2,013 | -2% | 0 | 0 | — |
case-12 | pass→pass | 20,869 | 3,110 | -85% | 1 | 1 | 0% | 2,415 | 841 | -65% | 0 | 0 | — |
case-13 | pass→pass | 18,218 | 14,389 | -21% | 1 | 1 | 0% | 2,168 | 1,914 | -12% | 0 | 0 | — |
case-14 | pass→pass | 23,515 | 12,097 | -49% | 1 | 1 | 0% | 3,002 | 2,914 | -3% | 0 | 0 | — |
case-15 | fail→fail | 11,529 | 7,469 | -35% | 1 | 1 | 0% | 1,808 | 754 | -58% | 0 | 0 | — |
case-16 | fail→pass | 19,453 | 3,571 | -82% | 1 | 1 | 0% | 2,272 | 937 | -59% | 0 | 0 | — |
case-17 | fail→pass | 11,584 | 8,083 | -30% | 1 | 1 | 0% | 1,145 | 781 | -32% | 0 | 0 | — |
case-18 | pass→pass | 13,060 | 12,748 | -2% | 1 | 1 | 0% | 1,629 | 1,639 | +1% | 0 | 0 | — |
case-19 | pass→pass | 18,428 | 18,684 | +1% | 1 | 1 | 0% | 1,957 | 2,317 | +18% | 0 | 0 | — |
case-20 | pass→fail | 24,227 | 36,873 | +52% | 1 | 1 | 0% | 4,331 | 6,297 | +45% | 0 | 0 | — |
case-21 | pass→pass | 16,915 | 9,011 | -47% | 1 | 1 | 0% | 2,040 | 1,863 | -9% | 0 | 0 | — |
case-22 | pass→pass | 15,814 | 16,314 | +3% | 1 | 1 | 0% | 2,081 | 2,385 | +15% | 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 19 counted toward the lift figure. The other 3 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 +18 percentage points is the difference between those two pass rates over the 19 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.