Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create visual assets (images, illustrations) using AI (OpenAI DALL-E).
.claude/skills/aiskillstore-asset-creator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -16% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 50% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -59% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 2% | 0% |
You are the Creative Director. You generate high-quality visual assets for the project using AI.
public/assets/ subfolders.Script: .claude/skills/asset-creator/scripts/generate-image.ts
Usage:
bashnpx tsx .claude/skills/asset-creator/scripts/generate-image.ts "<prompt>" "<filename>" "[folder]" "[style]"
Parameters:
prompt: Description of the image.filename: Name of the file (without extension).folder: Subfolder in public/ (default: assets/images).style: vivid or natural (default: vivid).Example:
bashnpx tsx .claude/skills/asset-creator/scripts/generate-image.ts "A futuristic dashboard on a laptop screen" "hero-dashboard" "assets/marketing" "natural"
When asked to "Create an image for X" or "Generate a hero image":
public folder.<Image src="/assets/marketing/hero-dashboard.png" ... />.For advanced configuration and prompt engineering tips, see reference.md.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-08 | fail→fail | 27,750 | 4,158 | -85% | 1 | 1 | 0% | 6,169 | 926 | -85% | 0 | 0 | — |
case-01 | fail→fail | 26,690 | 3,212 | -88% | 1 | 1 | 0% | 5,913 | 884 | -85% | 0 | 0 | — |
case-02 | fail→fail | 19,286 | 3,056 | -84% | 1 | 1 | 0% | 2,358 | 868 | -63% | 0 | 0 | — |
case-03 | fail→fail | 22,403 | 9,334 | -58% | 1 | 1 | 0% | 4,955 | 940 | -81% | 0 | 0 | — |
case-04 | fail→fail | 5,810 | 4,390 | -24% | 1 | 1 | 0% | 820 | 1,166 | +42% | 0 | 0 | — |
case-05 | fail→fail | 9,011 | 2,555 | -72% | 1 | 1 | 0% | 1,846 | 859 | -53% | 0 | 0 | — |
case-06 | fail→pass | 9,138 | 3,454 | -62% | 1 | 1 | 0% | 1,473 | 1,056 | -28% | 0 | 0 | — |
case-07 | fail→pass | 8,061 | 3,767 | -53% | 1 | 1 | 0% | 1,351 | 1,141 | -16% | 0 | 0 | — |
case-09 | fail→fail | 29,205 | 5,146 | -82% | 1 | 1 | 0% | 6,169 | 1,295 | -79% | 0 | 0 | — |
case-10 | fail→fail | 29,809 | 3,185 | -89% | 1 | 1 | 0% | 6,169 | 850 | -86% | 0 | 0 | — |
case-11 | fail→pass | 4,501 | 4,782 | +6% | 1 | 1 | 0% | 708 | 1,062 | +50% | 0 | 0 | — |
case-12 | pass→pass | 23,491 | 4,122 | -82% | 1 | 1 | 0% | 5,087 | 1,082 | -79% | 0 | 0 | — |
case-13 | fail→fail | 3,912 | 3,848 | -2% | 1 | 1 | 0% | 599 | 1,098 | +83% | 0 | 0 | — |
case-14 | fail→fail | 18,963 | 3,708 | -80% | 1 | 1 | 0% | 4,014 | 1,042 | -74% | 0 | 0 | — |
case-15 | fail→pass | 17,415 | 5,756 | -67% | 1 | 1 | 0% | 3,750 | 1,538 | -59% | 0 | 0 | — |
case-16 | fail→pass | 6,294 | 3,390 | -46% | 1 | 1 | 0% | 1,064 | 1,085 | +2% | 0 | 0 | — |
case-17 | pass→fail | 17,042 | 8,377 | -51% | 1 | 1 | 0% | 3,385 | 930 | -73% | 0 | 0 | — |
case-18 | pass→pass | 12,615 | 5,096 | -60% | 1 | 1 | 0% | 2,407 | 1,294 | -46% | 0 | 0 | — |
case-19 | fail→fail | 18,697 | 3,177 | -83% | 1 | 1 | 0% | 4,078 | 906 | -78% | 0 | 0 | — |
case-20 | pass→pass | 11,120 | 8,512 | -23% | 1 | 1 | 0% | 2,399 | 2,123 | -12% | 0 | 0 | — |
case-21 | pass→pass | 12,514 | 7,406 | -41% | 1 | 1 | 0% | 2,383 | 1,937 | -19% | 0 | 0 | — |
case-22 | pass→pass | 6,549 | 5,017 | -23% | 1 | 1 | 0% | 1,288 | 1,331 | +3% | 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 +18 percentage points is the difference between those two pass rates over the 20 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.