Install any skill in seconds. Free to start, no credit card required.
Get Started Free →HarmonyOS ArkTS rules for components, state, resources, layout, lifecycle, and accessibility
.claude/skills/amariahak-harmonyos-arkts-rules-for-components-state-resources-layout-lifecycle-and-a/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-15 | ✓→✓ | = Same ✓ | -31% | 0% |
| case-02 | ✓→✓ | = Same ✓ | -2% | 0% |
| case-05 | ✓→✓ | = Same ✓ | 17% | 0% |
HarmonyOS ArkTS rules for components, state, resources, layout, lifecycle, and accessibility
Synced from https://github.com/PatrickJS/awesome-cursorrules/tree/main/rules/harmony-arkts.mdc.
@Component for component definitions and PascalCase for component structs.build().build() last and keep it focused on UI composition.@State for component-owned state.@Prop for parent-to-child data.@Link only for intentional two-way binding.Column, Row, Stack, List, and other ArkUI primitives intentionally.$r() for app resources.build().console.log calls in production code.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-15 | pass→pass | 8,830 | 3,570 | -60% | 1 | 1 | 0% | 1,631 | 1,133 | -31% | 0 | 0 | — |
case-01 | fail→pass | 20,534 | 15,036 | -27% | 1 | 1 | 0% | 4,462 | 3,617 | -19% | 0 | 0 | — |
case-02 | pass→pass | 18,021 | 14,485 | -20% | 1 | 1 | 0% | 3,790 | 3,696 | -2% | 0 | 0 | — |
case-03 | fail→fail | 17,040 | 8,706 | -49% | 1 | 1 | 0% | 3,402 | 2,281 | -33% | 0 | 0 | — |
case-04 | fail→pass | 10,397 | 10,022 | -4% | 1 | 1 | 0% | 2,079 | 2,353 | +13% | 0 | 0 | — |
case-05 | pass→pass | 9,280 | 9,529 | +3% | 1 | 1 | 0% | 1,801 | 2,105 | +17% | 0 | 0 | — |
case-06 | pass→pass | 14,364 | 5,485 | -62% | 1 | 1 | 0% | 2,200 | 1,519 | -31% | 0 | 0 | — |
case-07 | pass→pass | 12,707 | 13,895 | +9% | 1 | 1 | 0% | 2,606 | 3,447 | +32% | 0 | 0 | — |
case-08 | pass→pass | 8,097 | 6,882 | -15% | 1 | 1 | 0% | 1,677 | 1,936 | +15% | 0 | 0 | — |
case-09 | pass→pass | 10,968 | 7,210 | -34% | 1 | 1 | 0% | 2,166 | 2,012 | -7% | 0 | 0 | — |
case-10 | pass→pass | 7,225 | 6,747 | -7% | 1 | 1 | 0% | 1,332 | 1,283 | -4% | 0 | 0 | — |
case-11 | pass→pass | 12,777 | 6,694 | -48% | 1 | 1 | 0% | 2,262 | 1,772 | -22% | 0 | 0 | — |
case-12 | pass→pass | 15,650 | 12,195 | -22% | 1 | 1 | 0% | 2,932 | 3,076 | +5% | 0 | 0 | — |
case-13 | pass→pass | 12,599 | 11,464 | -9% | 1 | 1 | 0% | 2,302 | 2,583 | +12% | 0 | 0 | — |
case-14 | pass→pass | 13,945 | 10,871 | -22% | 1 | 1 | 0% | 2,885 | 2,608 | -10% | 0 | 0 | — |
case-16 | pass→pass | 14,556 | 8,410 | -42% | 1 | 1 | 0% | 2,671 | 2,074 | -22% | 0 | 0 | — |
case-17 | pass→pass | 12,708 | 5,110 | -60% | 1 | 1 | 0% | 2,335 | 1,323 | -43% | 0 | 0 | — |
case-18 | pass→pass | 7,139 | 4,125 | -42% | 1 | 1 | 0% | 1,260 | 1,209 | -4% | 0 | 0 | — |
case-19 | pass→pass | 13,213 | 9,008 | -32% | 1 | 1 | 0% | 2,331 | 1,790 | -23% | 0 | 0 | — |
case-20 | fail→fail | 14,392 | 13,026 | -9% | 1 | 1 | 0% | 2,817 | 3,023 | +7% | 0 | 0 | — |
case-21 | pass→pass | 7,604 | 5,295 | -30% | 1 | 1 | 0% | 1,436 | 1,463 | +2% | 0 | 0 | — |
case-22 | pass→pass | 13,021 | 11,285 | -13% | 1 | 1 | 0% | 2,699 | 2,847 | +5% | 0 | 0 | — |
case-23 | pass→pass | 8,974 | 10,932 | +22% | 1 | 1 | 0% | 1,744 | 1,862 | +7% | 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. 23 cases were attempted. The headline lift of +9 percentage points is the difference between those two pass rates over the 23 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.