Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Route frontend work to the correct project UI, data-experience, motion, and polish skills. Use before designing or implementing a page, interaction, dashboard, mobile surface, or visual redesign so only compatible Profiles guide the work.
.claude/skills/qierkang-ui-experience-orchestrator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -64% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -16% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -27% | 0% |
Do not equate more animation or more Skills with better user experience.
Determine surface, stack, motion, phase, and artifact before selecting a Profile:
Run scripts/util-select-agent-profiles.sh with all five facts when available. Infer artifact from the requested deliverable: HTML mockup/clickable exploration means prototype, low-fidelity exploration means wireframe, token/component review means design-system, and changes to the real application mean implementation. If a fact cannot be derived, use the least invasive choice and state the assumption.
ui-prototype-html only.baoyu-design-prototyping for HTML design artifacts; never use its HTML or screenshots as formal raster assets or production implementation evidence.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 25,700 | 19,103 | -26% | 1 | 1 | 0% | 4,731 | 2,782 | -41% | 0 | 0 | — |
case-02 | fail→pass | 25,753 | 9,148 | -64% | 1 | 1 | 0% | 5,275 | 1,897 | -64% | 0 | 0 | — |
case-03 | fail→pass | 20,012 | 12,859 | -36% | 1 | 1 | 0% | 3,271 | 2,744 | -16% | 0 | 0 | — |
case-04 | pass→pass | 23,244 | 21,786 | -6% | 1 | 1 | 0% | 3,896 | 4,338 | +11% | 0 | 0 | — |
case-05 | pass→pass | 19,842 | 20,769 | +5% | 1 | 1 | 0% | 3,714 | 4,169 | +12% | 0 | 0 | — |
case-06 | pass→pass | 17,294 | 23,434 | +36% | 1 | 1 | 0% | 3,286 | 5,134 | +56% | 0 | 0 | — |
case-07 | fail→pass | 11,211 | 5,183 | -54% | 1 | 1 | 0% | 1,599 | 1,285 | -20% | 0 | 0 | — |
case-08 | fail→pass | 14,464 | 6,464 | -55% | 1 | 1 | 0% | 2,090 | 1,453 | -30% | 0 | 0 | — |
case-09 | pass→pass | 12,370 | 4,601 | -63% | 1 | 1 | 0% | 1,814 | 1,159 | -36% | 0 | 0 | — |
case-10 | pass→pass | 15,169 | 5,313 | -65% | 1 | 1 | 0% | 2,189 | 1,182 | -46% | 0 | 0 | — |
case-11 | pass→pass | 11,025 | 4,901 | -56% | 1 | 1 | 0% | 1,685 | 1,185 | -30% | 0 | 0 | — |
case-12 | pass→pass | 8,446 | 2,891 | -66% | 1 | 1 | 0% | 1,333 | 944 | -29% | 0 | 0 | — |
case-13 | pass→pass | 15,049 | 10,832 | -28% | 1 | 1 | 0% | 2,320 | 1,968 | -15% | 0 | 0 | — |
case-14 | fail→pass | 13,807 | 7,003 | -49% | 1 | 1 | 0% | 2,041 | 1,490 | -27% | 0 | 0 | — |
case-15 | pass→pass | 7,587 | 2,417 | -68% | 1 | 1 | 0% | 1,139 | 823 | -28% | 0 | 0 | — |
case-16 | fail→pass | 12,136 | 2,520 | -79% | 1 | 1 | 0% | 1,876 | 767 | -59% | 0 | 0 | — |
case-17 | pass→pass | 13,604 | 3,658 | -73% | 1 | 1 | 0% | 1,998 | 927 | -54% | 0 | 0 | — |
case-18 | fail→pass | 25,091 | 3,303 | -87% | 1 | 1 | 0% | 1,784 | 880 | -51% | 0 | 0 | — |
case-19 | fail→pass | 11,761 | 3,583 | -70% | 1 | 1 | 0% | 1,881 | 932 | -50% | 0 | 0 | — |
case-20 | pass→pass | 5,658 | 3,428 | -39% | 1 | 1 | 0% | 979 | 1,028 | +5% | 0 | 0 | — |
case-21 | fail→fail | 11,309 | 1,402 | -88% | 1 | 1 | 0% | 2,096 | 586 | -72% | 0 | 0 | — |
case-22 | fail→pass | 9,594 | 1,848 | -81% | 1 | 1 | 0% | 1,532 | 701 | -54% | 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 +41 percentage points is the difference between those two pass rates over the 21 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.