Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →실제 화면 캡처를 콘텐츠 이미지·GIF로 가공한다. 개인정보를 모자이크 처리하고, 강조 박스를 눈대중이 아니라 요소의 실제 렌더링 좌표를 측정해 배치한다.
.claude/skills/bam-bam-2-measured-ui-callouts/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -17% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 11% | 0% |
> 이 스킬이 시스템에 하는 일 (설치 전 확인) > > - 화면 캡처 이미지를 읽고 가공한 이미지를 새로 만듭니다. 원본은 수정하지 않습니다. > - 개인정보 영역을 모자이크·블러 처리하는 단계가 포함됩니다. 최종 확인은 사람이 하세요.
실제 화면을 장식용 이미지로 재제작하지 말고, 원본 화면을 증거 자료로 유지한다.
getBoundingClientRect()를 측정한다. 눈대중 좌표를 사용하지 않는다.clip으로 잘린 결과를 완성본으로 쓰지 않는다.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 14,639 | 27,180 | +86% | 1 | 1 | 0% | 2,200 | 5,631 | +156% | 0 | 0 | — |
case-02 | fail→pass | 17,714 | 16,259 | -8% | 1 | 1 | 0% | 2,500 | 3,122 | +25% | 0 | 0 | — |
case-03 | fail→fail | 20,819 | 16,116 | -23% | 1 | 1 | 0% | 3,157 | 2,740 | -13% | 0 | 0 | — |
case-04 | pass→pass | 15,151 | 13,436 | -11% | 1 | 1 | 0% | 2,369 | 2,715 | +15% | 0 | 0 | — |
case-05 | pass→pass | 19,054 | 15,143 | -21% | 1 | 1 | 0% | 3,169 | 3,156 | -0% | 0 | 0 | — |
case-06 | fail→pass | 27,656 | 8,247 | -70% | 1 | 1 | 0% | 2,154 | 1,992 | -8% | 0 | 0 | — |
case-07 | pass→pass | 13,712 | 11,372 | -17% | 1 | 1 | 0% | 1,981 | 2,187 | +10% | 0 | 0 | — |
case-08 | fail→pass | 18,630 | 10,720 | -42% | 1 | 1 | 0% | 2,625 | 2,185 | -17% | 0 | 0 | — |
case-09 | fail→pass | 15,234 | 10,598 | -30% | 1 | 1 | 0% | 2,016 | 2,046 | +1% | 0 | 0 | — |
case-10 | pass→pass | 13,797 | 9,600 | -30% | 1 | 1 | 0% | 1,906 | 2,041 | +7% | 0 | 0 | — |
case-11 | fail→pass | 14,531 | 10,961 | -25% | 1 | 1 | 0% | 2,100 | 2,334 | +11% | 0 | 0 | — |
case-12 | pass→pass | 12,546 | 8,706 | -31% | 1 | 1 | 0% | 1,796 | 2,175 | +21% | 0 | 0 | — |
case-13 | pass→pass | 20,444 | 11,443 | -44% | 1 | 1 | 0% | 2,126 | 2,399 | +13% | 0 | 0 | — |
case-14 | pass→pass | 15,519 | 9,936 | -36% | 1 | 1 | 0% | 2,168 | 2,109 | -3% | 0 | 0 | — |
case-15 | pass→pass | 12,728 | 13,369 | +5% | 1 | 1 | 0% | 1,883 | 2,174 | +15% | 0 | 0 | — |
case-16 | fail→pass | 14,949 | 9,559 | -36% | 1 | 1 | 0% | 2,024 | 2,349 | +16% | 0 | 0 | — |
case-17 | pass→pass | 13,512 | 9,199 | -32% | 1 | 1 | 0% | 2,378 | 2,019 | -15% | 0 | 0 | — |
case-18 | pass→pass | 14,459 | 8,690 | -40% | 1 | 1 | 0% | 1,871 | 1,894 | +1% | 0 | 0 | — |
case-19 | pass→pass | 21,195 | 18,767 | -11% | 1 | 1 | 0% | 3,281 | 3,630 | +11% | 0 | 0 | — |
case-20 | pass→pass | 18,981 | 18,601 | -2% | 1 | 1 | 0% | 3,027 | 3,827 | +26% | 0 | 0 | — |
case-21 | pass→pass | 20,358 | 19,517 | -4% | 1 | 1 | 0% | 3,640 | 4,127 | +13% | 0 | 0 | — |
case-22 | pass→pass | 17,334 | 22,101 | +28% | 1 | 1 | 0% | 3,570 | 4,906 | +37% | 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 +27 percentage points is the difference between those two pass rates over the 22 comparable cases.
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.