Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use this skill when reviewing a PR or a set of changes and you want a consistent, production-grade review pass: security first, then correctness, then performance, then style.
.claude/skills/amariahak-code-review/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-13 | ✓→✗ | ▼ Worse | 43% | 0% |
| case-18 | ✓→✓ | ＝ Same ✓ | 7% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 102% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | -5% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | -2% | 0% |
Use this skill when reviewing a PR or a set of changes and you want a consistent, production-grade review pass: security first, then correctness, then performance, then style.
Good review comments include:
When you see risk, ask for one of:
Prefer:
Avoid:
git diff and git log to understand scope and intent.read_file + grep to confirm a change doesn’t break callsites.grep/glob to find entry points and cross-file coupling when reviewing bigger refactors.Good reviewer flow:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-18 | pass→pass | 11,443 | 7,526 | -34% | 1 | 1 | 0% | 1,820 | 1,948 | +7% | 0 | 0 | — |
case-01 | fail→fail | 17,825 | 4,254 | -76% | 1 | 1 | 0% | 2,949 | 1,006 | -66% | 0 | 0 | — |
case-02 | fail→fail | 19,122 | 17,607 | -8% | 1 | 1 | 0% | 3,289 | 3,410 | +4% | 0 | 0 | — |
case-03 | fail→fail | 8,086 | 13,735 | +70% | 1 | 1 | 0% | 1,363 | 3,019 | +121% | 0 | 0 | — |
case-04 | pass→pass | 8,416 | 6,907 | -18% | 1 | 1 | 0% | 975 | 1,972 | +102% | 0 | 0 | — |
case-05 | pass→pass | 17,199 | 10,399 | -40% | 1 | 1 | 0% | 2,783 | 2,650 | -5% | 0 | 0 | — |
case-06 | pass→pass | 13,786 | 11,948 | -13% | 1 | 1 | 0% | 2,813 | 2,759 | -2% | 0 | 0 | — |
case-07 | pass→pass | 22,119 | 25,216 | +14% | 1 | 1 | 0% | 5,104 | 5,926 | +16% | 0 | 0 | — |
case-08 | pass→pass | 23,466 | 18,066 | -23% | 1 | 1 | 0% | 3,973 | 4,304 | +8% | 0 | 0 | — |
case-09 | pass→pass | 12,194 | 8,411 | -31% | 1 | 1 | 0% | 2,176 | 2,214 | +2% | 0 | 0 | — |
case-10 | pass→pass | 15,520 | 13,032 | -16% | 1 | 1 | 0% | 2,277 | 2,828 | +24% | 0 | 0 | — |
case-11 | pass→pass | 13,226 | 8,769 | -34% | 1 | 1 | 0% | 1,730 | 1,973 | +14% | 0 | 0 | — |
case-12 | pass→pass | 20,124 | 20,586 | +2% | 1 | 1 | 0% | 2,745 | 3,293 | +20% | 0 | 0 | — |
case-13 | pass→fail | 13,679 | 16,819 | +23% | 1 | 1 | 0% | 1,533 | 2,185 | +43% | 0 | 0 | — |
case-14 | pass→pass | 13,171 | 10,464 | -21% | 1 | 1 | 0% | 2,021 | 2,573 | +27% | 0 | 0 | — |
case-15 | pass→pass | 16,386 | 12,305 | -25% | 1 | 1 | 0% | 2,359 | 2,913 | +23% | 0 | 0 | — |
case-16 | pass→pass | 11,148 | 9,668 | -13% | 1 | 1 | 0% | 1,947 | 2,382 | +22% | 0 | 0 | — |
case-17 | pass→pass | 8,877 | 4,600 | -48% | 1 | 1 | 0% | 1,545 | 1,480 | -4% | 0 | 0 | — |
case-19 | pass→pass | 17,421 | 13,445 | -23% | 1 | 1 | 0% | 2,755 | 2,763 | +0% | 0 | 0 | — |
case-20 | pass→pass | 7,837 | 7,177 | -8% | 1 | 1 | 0% | 1,450 | 1,964 | +35% | 0 | 0 | — |
case-21 | pass→pass | 9,602 | 10,377 | +8% | 1 | 1 | 0% | 1,532 | 1,808 | +18% | 0 | 0 | — |
case-22 | pass→pass | 15,330 | 6,892 | -55% | 1 | 1 | 0% | 2,461 | 1,788 | -27% | 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 21 counted toward the lift figure. The other 1 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 -100 percentage points is the difference between those two pass rates over the 21 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Other measured skills in the registry, with their headline benchmark lift.