Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when reviewing rendered HTML, interactive components, or design-system patterns related to Avoid sensory-only instructions. Check native semantics first, then inspect keyboard behavior, focus flow, accessible names, and screen-reader output where relevant.
.claude/skills/thedaviddias-sensory-instructions/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-18 | ✗→✓ | ▲ Improved | -27% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | 64% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | 16% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 30% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 36% | 0% |
Color-blind users can't see 'click the red button,' screen reader users can't perceive 'the menu on the right,' and deaf users miss audio cues—multi-modal instructions include everyone.
Review instructions and verify they don't rely only on color ('click the red button'), shape ('the circular icon'), location ('the menu on the right'), or sound ('when you hear the beep'). Ensure text labels supplement visual cues.
Add text labels to visual instructions. Instead of 'click the green button', use 'click the Submit button (green)'. Combine multiple cues so users who cannot perceive one can still understand via another.
Explain how color-blind users cannot distinguish color-only cues, screen reader users cannot perceive location or shape, and deaf users miss audio-only signals, making multi-modal instructions essential.
Review the rendered markup and interactive states that affect Avoid sensory-only instructions. Flag exact elements, roles, labels, focus behavior, or keyboard interactions that violate the rule, and note how to verify the fix with browser accessibility tooling or assistive tech.
For full implementation details, code examples, and framework-specific guidance, see references/rule.md.
Rule page: https://frontendchecklist.io/en/rules/accessibility/sensory-instructions
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 3,849 | 4,203 | +9% | 1 | 1 | 0% | 716 | 1,176 | +64% | 0 | 0 | — |
case-02 | pass→pass | 9,772 | 9,642 | -1% | 1 | 1 | 0% | 1,741 | 2,024 | +16% | 0 | 0 | — |
case-03 | pass→pass | 8,000 | 9,218 | +15% | 1 | 1 | 0% | 1,594 | 2,076 | +30% | 0 | 0 | — |
case-04 | pass→pass | 6,678 | 7,221 | +8% | 1 | 1 | 0% | 1,185 | 1,610 | +36% | 0 | 0 | — |
case-05 | pass→pass | 6,232 | 7,257 | +16% | 1 | 1 | 0% | 1,170 | 1,599 | +37% | 0 | 0 | — |
case-06 | pass→pass | 7,142 | 4,946 | -31% | 1 | 1 | 0% | 1,292 | 1,129 | -13% | 0 | 0 | — |
case-07 | pass→pass | 7,299 | 5,921 | -19% | 1 | 1 | 0% | 1,200 | 1,335 | +11% | 0 | 0 | — |
case-08 | pass→pass | 9,902 | 7,421 | -25% | 1 | 1 | 0% | 1,738 | 1,649 | -5% | 0 | 0 | — |
case-09 | pass→pass | 10,546 | 7,621 | -28% | 1 | 1 | 0% | 1,802 | 1,557 | -14% | 0 | 0 | — |
case-10 | pass→pass | 12,188 | 9,303 | -24% | 1 | 1 | 0% | 1,977 | 1,875 | -5% | 0 | 0 | — |
case-11 | pass→pass | 8,234 | 7,279 | -12% | 1 | 1 | 0% | 1,703 | 1,725 | +1% | 0 | 0 | — |
case-12 | pass→pass | 6,154 | 4,244 | -31% | 1 | 1 | 0% | 1,087 | 1,073 | -1% | 0 | 0 | — |
case-13 | pass→pass | 8,180 | 8,358 | +2% | 1 | 1 | 0% | 1,549 | 1,810 | +17% | 0 | 0 | — |
case-14 | pass→pass | 14,461 | 12,315 | -15% | 1 | 1 | 0% | 2,544 | 2,757 | +8% | 0 | 0 | — |
case-15 | pass→pass | 8,419 | 6,029 | -28% | 1 | 1 | 0% | 1,439 | 1,272 | -12% | 0 | 0 | — |
case-16 | pass→pass | 8,197 | 8,861 | +8% | 1 | 1 | 0% | 1,598 | 2,099 | +31% | 0 | 0 | — |
case-17 | pass→pass | 10,258 | 8,115 | -21% | 1 | 1 | 0% | 1,715 | 1,708 | -0% | 0 | 0 | — |
case-18 | fail→pass | 10,003 | 4,759 | -52% | 1 | 1 | 0% | 1,658 | 1,213 | -27% | 0 | 0 | — |
case-19 | pass→pass | 6,574 | 5,636 | -14% | 1 | 1 | 0% | 1,158 | 1,196 | +3% | 0 | 0 | — |
case-20 | pass→pass | 6,629 | 4,745 | -28% | 1 | 1 | 0% | 1,166 | 1,076 | -8% | 0 | 0 | — |
case-21 | pass→pass | 8,890 | 7,467 | -16% | 1 | 1 | 0% | 1,593 | 1,717 | +8% | 0 | 0 | — |
case-22 | pass→pass | 9,621 | 6,626 | -31% | 1 | 1 | 0% | 1,579 | 1,419 | -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. 22 cases were attempted. The headline lift of +5 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.