Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Review code changes for bugs, security issues, performance problems, and style violations. Use when asked to review code, a PR, or recent changes.
.claude/skills/owainlewis-code-review/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✓→✗ | ▼ Worse | 31% | 0% |
| case-16 | ✓→✗ | ▼ Worse | -48% | 0% |
| case-20 | ✓→✗ | ▼ Worse | -13% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | -29% | 0% |
| case-07 | ✓→✓ | ＝ Same ✓ | -29% | 0% |
When reviewing code, follow this process:
Run git diff (or git diff --staged for staged changes) to see what changed. Read the modified files in full to understand the surrounding context.
Review each changed file for:
Correctness
Security
Performance
Maintainability
Format your review as:
Be direct. Don't pad with praise. Focus on things that matter.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 3,235 | 4,235 | +31% | 1 | 1 | 0% | 472 | 396 | -16% | 0 | 0 | — |
case-02 | fail→fail | 4,286 | 2,822 | -34% | 1 | 1 | 0% | 607 | 508 | -16% | 0 | 0 | — |
case-03 | pass→fail | 7,117 | 7,594 | +7% | 1 | 1 | 0% | 1,156 | 1,514 | +31% | 0 | 0 | — |
case-04 | fail→fail | 5,828 | 2,498 | -57% | 1 | 1 | 0% | 875 | 561 | -36% | 0 | 0 | — |
case-05 | pass→pass | 11,428 | 7,718 | -32% | 1 | 1 | 0% | 2,101 | 1,486 | -29% | 0 | 0 | — |
case-06 | fail→fail | 3,740 | 3,986 | +7% | 1 | 1 | 0% | 480 | 382 | -20% | 0 | 0 | — |
case-07 | pass→pass | 13,763 | 7,797 | -43% | 1 | 1 | 0% | 1,612 | 1,143 | -29% | 0 | 0 | — |
case-08 | pass→pass | 10,972 | 5,060 | -54% | 1 | 1 | 0% | 1,777 | 1,058 | -40% | 0 | 0 | — |
case-09 | pass→pass | 10,420 | 5,432 | -48% | 1 | 1 | 0% | 1,802 | 1,002 | -44% | 0 | 0 | — |
case-10 | pass→pass | 2,819 | 6,731 | +139% | 1 | 1 | 0% | 401 | 1,384 | +245% | 0 | 0 | — |
case-11 | fail→fail | 2,211 | 2,783 | +26% | 1 | 1 | 0% | 276 | 590 | +114% | 0 | 0 | — |
case-12 | pass→pass | 25,716 | 6,729 | -74% | 1 | 1 | 0% | 1,131 | 1,325 | +17% | 0 | 0 | — |
case-13 | pass→pass | 6,741 | 3,790 | -44% | 1 | 1 | 0% | 1,173 | 889 | -24% | 0 | 0 | — |
case-14 | pass→pass | 10,624 | 4,890 | -54% | 1 | 1 | 0% | 1,648 | 947 | -43% | 0 | 0 | — |
case-15 | pass→pass | 9,269 | 12,026 | +30% | 1 | 1 | 0% | 945 | 1,281 | +36% | 0 | 0 | — |
case-16 | pass→fail | 16,383 | 9,994 | -39% | 1 | 1 | 0% | 2,494 | 1,295 | -48% | 0 | 0 | — |
case-17 | pass→pass | 14,016 | 6,271 | -55% | 1 | 1 | 0% | 2,313 | 1,298 | -44% | 0 | 0 | — |
case-18 | pass→pass | 11,354 | 6,189 | -45% | 1 | 1 | 0% | 1,668 | 1,141 | -32% | 0 | 0 | — |
case-19 | pass→pass | 9,494 | 5,359 | -44% | 1 | 1 | 0% | 1,669 | 1,149 | -31% | 0 | 0 | — |
case-20 | pass→fail | 14,877 | 19,970 | +34% | 1 | 1 | 0% | 2,026 | 1,756 | -13% | 0 | 0 | — |
case-21 | pass→pass | 11,933 | 6,842 | -43% | 1 | 1 | 0% | 2,090 | 1,416 | -32% | 0 | 0 | — |
case-22 | pass→pass | 12,291 | 5,907 | -52% | 1 | 1 | 0% | 1,971 | 1,171 | -41% | 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 20 counted toward the lift figure. The other 2 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 20 comparable cases. 3 cases got worse with the skill loaded, and they are included in that figure.
Other measured skills in the registry, with their headline benchmark lift.