Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Develop native iOS applications with Swift/SwiftUI. Masters iOS 18, SwiftUI, UIKit integration, Core Data, networking, and App Store optimization.
.claude/skills/sickn33-ios-developer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-22 | ✗→✓ | ▲ Improved | 82% | 0% |
| case-25 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-23 | ✓→✗ | ▼ Worse | -13% | 0% |
| case-05 | ✓→✓ | = Same ✓ | 67% | 0% |
| case-10 | ✓→✓ | = Same ✓ | 48% | 0% |
resources/implementation-playbook.md.You are an iOS development expert specializing in native iOS app development with comprehensive knowledge of the Apple ecosystem.
Expert iOS developer specializing in Swift 6, SwiftUI, and native iOS application development. Masters modern iOS architecture patterns, performance optimization, and Apple platform integrations while maintaining code quality and App Store compliance.
Focus on Swift-first solutions with modern iOS patterns. Include comprehensive error handling, accessibility support, and App Store compliance considerations.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | fail→fail | 16,098 | 20,287 | +26% | 1 | 1 | 0% | 2,840 | 5,772 | +103% | 0 | 0 | — |
case-02 | fail→fail | 28,054 | 24,517 | -13% | 1 | 1 | 0% | 4,769 | 6,124 | +28% | 0 | 0 | — |
case-01 | fail→fail | 32,639 | 28,855 | -12% | 1 | 1 | 0% | 5,772 | 6,553 | +14% | 0 | 0 | — |
case-03 | fail→fail | 29,057 | 28,074 | -3% | 1 | 1 | 0% | 4,878 | 6,423 | +32% | 0 | 0 | — |
case-05 | pass→pass | 16,628 | 17,248 | +4% | 1 | 1 | 0% | 2,962 | 4,933 | +67% | 0 | 0 | — |
case-06 | fail→fail | 17,578 | 15,965 | -9% | 1 | 1 | 0% | 3,001 | 4,720 | +57% | 0 | 0 | — |
case-07 | fail→fail | 16,316 | 14,211 | -13% | 1 | 1 | 0% | 2,653 | 4,129 | +56% | 0 | 0 | — |
case-08 | fail→fail | 10,537 | 13,169 | +25% | 1 | 1 | 0% | 1,923 | 3,972 | +107% | 0 | 0 | — |
case-09 | fail→fail | 13,687 | 13,659 | -0% | 1 | 1 | 0% | 1,988 | 3,580 | +80% | 0 | 0 | — |
case-10 | pass→pass | 16,593 | 16,402 | -1% | 1 | 1 | 0% | 3,098 | 4,591 | +48% | 0 | 0 | — |
case-11 | fail→fail | 14,584 | 15,379 | +5% | 1 | 1 | 0% | 2,083 | 4,086 | +96% | 0 | 0 | — |
case-12 | fail→fail | 16,823 | 15,407 | -8% | 1 | 1 | 0% | 2,477 | 4,181 | +69% | 0 | 0 | — |
case-13 | pass→pass | 17,895 | 17,932 | +0% | 1 | 1 | 0% | 2,949 | 4,709 | +60% | 0 | 0 | — |
case-14 | fail→fail | 14,862 | 17,991 | +21% | 1 | 1 | 0% | 2,252 | 4,624 | +105% | 0 | 0 | — |
case-15 | pass→pass | 13,616 | 12,937 | -5% | 1 | 1 | 0% | 2,218 | 3,801 | +71% | 0 | 0 | — |
case-16 | fail→fail | 31,619 | 24,162 | -24% | 1 | 1 | 0% | 6,057 | 5,949 | -2% | 0 | 0 | — |
case-17 | pass→pass | 16,248 | 15,184 | -7% | 1 | 1 | 0% | 2,334 | 3,941 | +69% | 0 | 0 | — |
case-18 | fail→fail | 15,733 | 16,482 | +5% | 1 | 1 | 0% | 2,158 | 4,361 | +102% | 0 | 0 | — |
case-19 | pass→pass | 15,367 | 16,384 | +7% | 1 | 1 | 0% | 2,551 | 4,568 | +79% | 0 | 0 | — |
case-20 | fail→fail | 20,846 | 21,721 | +4% | 1 | 1 | 0% | 3,056 | 5,174 | +69% | 0 | 0 | — |
case-21 | fail→fail | 16,544 | 16,544 | 0% | 1 | 1 | 0% | 2,306 | 4,197 | +82% | 0 | 0 | — |
case-22 | fail→pass | 13,545 | 15,785 | +17% | 1 | 1 | 0% | 2,267 | 4,136 | +82% | 0 | 0 | — |
case-23 | pass→fail | 17,017 | 7,347 | -57% | 1 | 1 | 0% | 3,425 | 2,985 | -13% | 0 | 0 | — |
case-24 | fail→fail | 23,360 | 19,775 | -15% | 1 | 1 | 0% | 4,699 | 5,206 | +11% | 0 | 0 | — |
case-25 | fail→pass | 12,969 | 9,645 | -26% | 1 | 1 | 0% | 2,637 | 3,529 | +34% | 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. 25 cases were attempted. The headline lift of +4 percentage points is the difference between those two pass rates over the 25 comparable cases. 1 case got worse with the skill loaded, and it is 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.