Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Cursor rules for avoiding placeholder or broken image URLs, using stable visual assets, and checking alt text, dimensions, licensing, and responsive behavior.
.claude/skills/amariahak-cursor-rules-for-avoiding-placeholder-or-broken-image-urls-using-stable-visual/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | 6% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -2% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -19% | 0% |
Cursor rules for avoiding placeholder or broken image URLs, using stable visual assets, and checking alt text, dimensions, licensing, and responsive behavior.
Synced from https://github.com/PatrickJS/awesome-cursorrules/tree/main/rules/landing-page-image-quality-cursorrules-prompt-file.mdc.
When generating or editing user-facing pages, use real, relevant visual assets instead of placeholder image services.
placehold.co, via.placeholder.com, dummyimage.com, picsum.photos, random image URLs, or broken/deprecated image services.width and height attributes or CSS aspect-ratio constraints to prevent layout shift.alt text that describes the image's purpose in context.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | fail→pass | 17,154 | 16,661 | -3% | 1 | 1 | 0% | 3,640 | 3,870 | +6% | 0 | 0 | — |
case-01 | fail→fail | 24,526 | 17,282 | -30% | 1 | 1 | 0% | 3,866 | 3,876 | +0% | 0 | 0 | — |
case-02 | fail→pass | 21,434 | 20,884 | -3% | 1 | 1 | 0% | 4,512 | 4,445 | -1% | 0 | 0 | — |
case-03 | fail→fail | 26,330 | 16,584 | -37% | 1 | 1 | 0% | 3,627 | 4,114 | +13% | 0 | 0 | — |
case-04 | pass→pass | 13,471 | 16,266 | +21% | 1 | 1 | 0% | 2,831 | 3,966 | +40% | 0 | 0 | — |
case-05 | pass→pass | 8,201 | 4,868 | -41% | 1 | 1 | 0% | 1,264 | 1,239 | -2% | 0 | 0 | — |
case-06 | pass→pass | 19,911 | 14,474 | -27% | 1 | 1 | 0% | 3,167 | 3,559 | +12% | 0 | 0 | — |
case-07 | pass→pass | 12,495 | 9,395 | -25% | 1 | 1 | 0% | 2,274 | 2,191 | -4% | 0 | 0 | — |
case-08 | fail→pass | 28,560 | 19,375 | -32% | 1 | 1 | 0% | 4,816 | 4,916 | +2% | 0 | 0 | — |
case-10 | fail→pass | 14,854 | 13,745 | -7% | 1 | 1 | 0% | 3,469 | 3,393 | -2% | 0 | 0 | — |
case-11 | fail→pass | 19,805 | 15,888 | -20% | 1 | 1 | 0% | 4,899 | 3,980 | -19% | 0 | 0 | — |
case-12 | pass→pass | 11,847 | 11,139 | -6% | 1 | 1 | 0% | 2,553 | 2,489 | -3% | 0 | 0 | — |
case-13 | pass→pass | 14,154 | 13,970 | -1% | 1 | 1 | 0% | 2,737 | 3,499 | +28% | 0 | 0 | — |
case-14 | fail→fail | 13,417 | 16,796 | +25% | 1 | 1 | 0% | 3,184 | 2,848 | -11% | 0 | 0 | — |
case-15 | pass→pass | 17,420 | 17,653 | +1% | 1 | 1 | 0% | 3,610 | 4,092 | +13% | 0 | 0 | — |
case-16 | fail→pass | 15,496 | 16,931 | +9% | 1 | 1 | 0% | 3,218 | 3,976 | +24% | 0 | 0 | — |
case-17 | fail→fail | 18,332 | 14,862 | -19% | 1 | 1 | 0% | 3,974 | 3,539 | -11% | 0 | 0 | — |
case-18 | pass→pass | 16,112 | 22,943 | +42% | 1 | 1 | 0% | 2,308 | 2,838 | +23% | 0 | 0 | — |
case-19 | pass→pass | 23,877 | 21,799 | -9% | 1 | 1 | 0% | 2,978 | 5,345 | +79% | 0 | 0 | — |
case-20 | pass→pass | 15,709 | 16,959 | +8% | 1 | 1 | 0% | 3,328 | 3,825 | +15% | 0 | 0 | — |
case-21 | pass→pass | 18,702 | 10,302 | -45% | 1 | 1 | 0% | 2,410 | 2,456 | +2% | 0 | 0 | — |
case-22 | pass→pass | 14,106 | 11,269 | -20% | 1 | 1 | 0% | 2,631 | 2,391 | -9% | 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.