Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Review generated code against the PRD, tech design, and verification checklist.
.claude/skills/khazp-vibe-review/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -67% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -60% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 19% | 0% |
| case-08 | ✓→✗ | ▼ Worse | 24% | 0% |
| case-11 | ✓→✗ | ▼ Worse | 90% | 0% |
Read the actual diff, applicable instructions, acceptance criteria, and check results. Trace changed behavior into its callers and tests. Look for concrete regressions, missing error handling, incorrect assumptions, and relevant authorization or data exposure issues.
Report actionable findings with file/line, trigger, impact, and suggested correction. Distinguish demonstrated defects from missing evidence. Prioritize correctness and scope over cosmetic preferences. Do not claim independent checks unless you ran them. If there are no findings, say so and disclose remaining verification gaps. Review does not authorize unrelated edits or publication.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 11,967 | 43,696 | +265% | 1 | 1 | 0% | 1,021 | 2,216 | +117% | 0 | 0 | — |
case-02 | fail→fail | 17,117 | 7,419 | -57% | 1 | 1 | 0% | 803 | 784 | -2% | 0 | 0 | — |
case-03 | fail→pass | 30,397 | 6,128 | -80% | 1 | 1 | 0% | 2,899 | 950 | -67% | 0 | 0 | — |
case-04 | pass→pass | 6,927 | 5,939 | -14% | 1 | 1 | 0% | 613 | 915 | +49% | 0 | 0 | — |
case-05 | fail→pass | 14,810 | 8,117 | -45% | 1 | 1 | 0% | 2,786 | 1,126 | -60% | 0 | 0 | — |
case-06 | fail→fail | 30,115 | 39,279 | +30% | 1 | 1 | 0% | 1,048 | 2,019 | +93% | 0 | 0 | — |
case-07 | pass→pass | 16,843 | 29,474 | +75% | 1 | 1 | 0% | 2,660 | 2,776 | +4% | 0 | 0 | — |
case-08 | pass→fail | 8,057 | 11,238 | +39% | 1 | 1 | 0% | 1,123 | 1,387 | +24% | 0 | 0 | — |
case-09 | pass→pass | 5,118 | 3,396 | -34% | 1 | 1 | 0% | 669 | 625 | -7% | 0 | 0 | — |
case-10 | fail→pass | 20,349 | 50,261 | +147% | 1 | 1 | 0% | 1,865 | 2,228 | +19% | 0 | 0 | — |
case-11 | pass→fail | 41,587 | 25,605 | -38% | 1 | 1 | 0% | 1,432 | 2,720 | +90% | 0 | 0 | — |
case-12 | pass→pass | 13,558 | 17,410 | +28% | 1 | 1 | 0% | 2,006 | 2,682 | +34% | 0 | 0 | — |
case-13 | pass→pass | 11,363 | 13,138 | +16% | 1 | 1 | 0% | 1,558 | 1,433 | -8% | 0 | 0 | — |
case-14 | pass→pass | 9,949 | 15,717 | +58% | 1 | 1 | 0% | 1,539 | 2,003 | +30% | 0 | 0 | — |
case-15 | pass→pass | 6,730 | 8,403 | +25% | 1 | 1 | 0% | 859 | 1,410 | +64% | 0 | 0 | — |
case-16 | pass→pass | 8,528 | 7,622 | -11% | 1 | 1 | 0% | 1,496 | 1,389 | -7% | 0 | 0 | — |
case-17 | pass→pass | 14,186 | 20,168 | +42% | 1 | 1 | 0% | 2,140 | 2,128 | -1% | 0 | 0 | — |
case-18 | pass→pass | 10,241 | 10,588 | +3% | 1 | 1 | 0% | 1,814 | 1,822 | +0% | 0 | 0 | — |
case-19 | pass→pass | 10,564 | 17,111 | +62% | 1 | 1 | 0% | 1,894 | 2,597 | +37% | 0 | 0 | — |
case-20 | pass→pass | 9,543 | 12,829 | +34% | 1 | 1 | 0% | 1,744 | 2,126 | +22% | 0 | 0 | — |
case-21 | pass→pass | 12,470 | 18,041 | +45% | 1 | 1 | 0% | 1,818 | 2,057 | +13% | 0 | 0 | — |
case-22 | pass→pass | 12,532 | 14,827 | +18% | 1 | 1 | 0% | 1,878 | 1,683 | -10% | 0 | 0 | — |
case-23 | pass→pass | 69,393 | 14,037 | -80% | 1 | 1 | 0% | 2,547 | 2,303 | -10% | 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. 23 cases were attempted. The headline lift of +4 percentage points is the difference between those two pass rates over the 23 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.