Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Mobile development knowledge reference covering iOS (SwiftUI), Android (Jetpack Compose), React Native, and Flutter. Use when building mobile apps, working with cross-platform frameworks, or implementing native UI patterns.
.claude/skills/telagod-developing-mobile-apps/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 18% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -57% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -70% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | -1% | 0% |
> 判断先于执行：决定「是否做 / 选什么 / 如何取舍」（栈、方案、架构、权衡）前，先读领域判断内核 skills/_kernel/frontend/SKILL.md——它管 judgment，本秘典管 execution；冲突时以内核判断为准。
原生：iOS(SwiftUI/UIKit) | Android(Compose/Kotlin)
跨平台：React Native(TS) | Flutter(Dart)SwiftUI 优先 | @MainActor 线程安全 | async/await | 依赖注入 | LazyVStack | Keychain 存敏感 | ViewModel 单测+Mock
Compose 优先 | StateFlow 替代 LiveData | Hilt 注入 | Room 持久化 | key 优化 LazyColumn | remember 防重组 | ViewModel 单测(runTest)
列表优化(FlatList/ListView.builder+key) | 状态管理(RTK/Riverpod) | 原生桥接验证 | 冷启动<1.5s | 渲染>55fps
Web 背景→RN | 极致动画/UI 定制→Flutter | 大量原生交互→RN | 极致原生体验→原生
SwiftUI/Compose/RN/Flutter API 详情详见 references/details.md
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 21,713 | 21,594 | -1% | 1 | 1 | 0% | 3,344 | 3,879 | +16% | 0 | 0 | — |
case-02 | fail→pass | 23,830 | 25,909 | +9% | 1 | 1 | 0% | 3,455 | 4,079 | +18% | 0 | 0 | — |
case-03 | pass→pass | 18,146 | 16,561 | -9% | 1 | 1 | 0% | 2,852 | 2,816 | -1% | 0 | 0 | — |
case-04 | pass→pass | 20,935 | 17,698 | -15% | 1 | 1 | 0% | 2,915 | 3,083 | +6% | 0 | 0 | — |
case-05 | fail→pass | 16,333 | 4,867 | -70% | 1 | 1 | 0% | 2,386 | 1,017 | -57% | 0 | 0 | — |
case-06 | pass→pass | 22,454 | 17,507 | -22% | 1 | 1 | 0% | 3,016 | 3,246 | +8% | 0 | 0 | — |
case-07 | pass→pass | 16,070 | 13,083 | -19% | 1 | 1 | 0% | 2,395 | 2,351 | -2% | 0 | 0 | — |
case-08 | pass→pass | 12,704 | 3,462 | -73% | 1 | 1 | 0% | 1,947 | 855 | -56% | 0 | 0 | — |
case-09 | pass→pass | 7,634 | 6,154 | -19% | 1 | 1 | 0% | 1,143 | 1,136 | -1% | 0 | 0 | — |
case-10 | pass→pass | 20,504 | 19,986 | -3% | 1 | 1 | 0% | 2,998 | 3,272 | +9% | 0 | 0 | — |
case-11 | pass→pass | 21,945 | 19,134 | -13% | 1 | 1 | 0% | 3,072 | 2,984 | -3% | 0 | 0 | — |
case-12 | pass→pass | 18,083 | 8,480 | -53% | 1 | 1 | 0% | 2,400 | 1,523 | -37% | 0 | 0 | — |
case-13 | pass→pass | 18,368 | 11,791 | -36% | 1 | 1 | 0% | 2,661 | 1,994 | -25% | 0 | 0 | — |
case-14 | fail→pass | 19,302 | 15,876 | -18% | 1 | 1 | 0% | 2,783 | 2,553 | -8% | 0 | 0 | — |
case-15 | pass→pass | 20,743 | 10,994 | -47% | 1 | 1 | 0% | 2,979 | 2,000 | -33% | 0 | 0 | — |
case-16 | pass→pass | 20,347 | 21,790 | +7% | 1 | 1 | 0% | 2,822 | 3,159 | +12% | 0 | 0 | — |
case-17 | pass→pass | 15,638 | 4,964 | -68% | 1 | 1 | 0% | 2,154 | 1,025 | -52% | 0 | 0 | — |
case-18 | fail→pass | 16,520 | 2,984 | -82% | 1 | 1 | 0% | 2,418 | 735 | -70% | 0 | 0 | — |
case-19 | pass→pass | 17,325 | 12,341 | -29% | 1 | 1 | 0% | 2,379 | 1,976 | -17% | 0 | 0 | — |
case-20 | pass→pass | 21,448 | 19,745 | -8% | 1 | 1 | 0% | 3,232 | 3,758 | +16% | 0 | 0 | — |
case-21 | pass→pass | 18,845 | 20,043 | +6% | 1 | 1 | 0% | 3,018 | 3,658 | +21% | 0 | 0 | — |
case-22 | pass→pass | 20,987 | 31,586 | +51% | 1 | 1 | 0% | 3,173 | 4,221 | +33% | 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 +18 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.