Install any skill in seconds. Free to start, no credit card required.
Get Started Free →iPhone + MacBook 仿 GLTF 静态展架, 屏幕内嵌真实 HTML 内容, 玻璃镜头折射, 360° 转盘构图
.claude/skills/nexu-io-mockup-device-3d/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | 12% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 181% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 246% | 0% |
| case-19 | ✗→✓ | ▲ Improved | 251% | 0% |
【模板: 设备 3D 展架 (Device 3D Showcase / HTML-in-Canvas)】 【意图】产品发布、App 演示、设计稿展示。把用户提供的 UI 内容真实渲染到 iPhone / MacBook "屏幕"里, 周围用 CSS 3D transform 模拟 GLTF 模型的玻璃 / 高光 / 折射。Inspired by hyperframes vfx-iphone-device。
【硬性构图】
radial-gradient(#1a1a1f → #0a0a0f), 底部反射地面 (mirror gradient)。transform: rotateY(-12deg) rotateX(4deg) translateZ(40px); 边框钛金属银 #a8a8ad (实心 4px) + 屏幕圆角 56px; 屏幕内嵌 iframe-like div, 真实渲染用户的 HTML 内容 (mobile viewport 375×812)。rotateY(8deg); 上盖屏幕嵌入桌面 viewport 内容 (1440×900 缩放); 底座键盘 + trackpad 用 CSS 阴影线条绘制 (不画键帽细节)。radial-gradient(ellipse, rgba(255,255,255,0.4) 0%, transparent 60%) 的椭圆 highlight, 模拟 morphing glass lens。transform: scaleY(-1) + mask-image: linear-gradient(to bottom, rgba(0,0,0,0.4), transparent 70%)。【屏幕内容来源】
【可选附加元素】
@keyframes turntable rotateY -12 ↔ 12, ease-in-out infinite alternate; 可被 prefers-reduced-motion 关闭。【设计细节】
Inter Tight / SF Pro 风格; 设备内根据用户内容自适应。<div class="screen"> + Tailwind 渲染内容。| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | fail→fail | 29,949 | 30,174 | +1% | 1 | 1 | 0% | 6,218 | 6,989 | +12% | 0 | 0 | — |
case-01 | fail→fail | 29,972 | 29,611 | -1% | 1 | 1 | 0% | 6,236 | 7,007 | +12% | 0 | 0 | — |
case-03 | fail→fail | 29,618 | 29,554 | -0% | 1 | 1 | 0% | 6,214 | 6,985 | +12% | 0 | 0 | — |
case-04 | pass→pass | 13,182 | 29,525 | +124% | 1 | 1 | 0% | 2,153 | 6,955 | +223% | 0 | 0 | — |
case-05 | pass→pass | 15,651 | 32,126 | +105% | 1 | 1 | 0% | 2,664 | 6,952 | +161% | 0 | 0 | — |
case-06 | fail→pass | 25,736 | 27,783 | +8% | 1 | 1 | 0% | 6,188 | 6,959 | +12% | 0 | 0 | — |
case-07 | fail→fail | 14,960 | 28,254 | +89% | 1 | 1 | 0% | 3,089 | 6,870 | +122% | 0 | 0 | — |
case-08 | fail→fail | 26,931 | 28,092 | +4% | 1 | 1 | 0% | 5,065 | 6,948 | +37% | 0 | 0 | — |
case-09 | fail→fail | 13,295 | 29,180 | +119% | 1 | 1 | 0% | 2,569 | 6,942 | +170% | 0 | 0 | — |
case-10 | fail→fail | 27,646 | 31,318 | +13% | 1 | 1 | 0% | 6,171 | 6,943 | +13% | 0 | 0 | — |
case-11 | fail→pass | 13,515 | 8,469 | -37% | 1 | 1 | 0% | 2,630 | 2,358 | -10% | 0 | 0 | — |
case-12 | fail→fail | 13,344 | 29,374 | +120% | 1 | 1 | 0% | 2,234 | 6,943 | +211% | 0 | 0 | — |
case-13 | fail→pass | 15,076 | 26,028 | +73% | 1 | 1 | 0% | 2,469 | 6,934 | +181% | 0 | 0 | — |
case-14 | pass→pass | 22,434 | 28,713 | +28% | 1 | 1 | 0% | 3,957 | 6,932 | +75% | 0 | 0 | — |
case-15 | fail→fail | 29,135 | 28,042 | -4% | 1 | 1 | 0% | 6,192 | 6,963 | +12% | 0 | 0 | — |
case-16 | pass→pass | 12,828 | 28,931 | +126% | 1 | 1 | 0% | 2,455 | 6,965 | +184% | 0 | 0 | — |
case-17 | fail→fail | 14,280 | 29,987 | +110% | 1 | 1 | 0% | 2,570 | 6,958 | +171% | 0 | 0 | — |
case-18 | fail→pass | 10,613 | 27,508 | +159% | 1 | 1 | 0% | 2,006 | 6,934 | +246% | 0 | 0 | — |
case-19 | fail→pass | 10,890 | 28,017 | +157% | 1 | 1 | 0% | 1,979 | 6,942 | +251% | 0 | 0 | — |
case-20 | pass→fail | 22,398 | 27,305 | +22% | 1 | 1 | 0% | 4,898 | 6,984 | +43% | 0 | 0 | — |
case-21 | pass→pass | 25,635 | 27,068 | +6% | 1 | 1 | 0% | 6,065 | 6,964 | +15% | 0 | 0 | — |
case-22 | fail→fail | 9,639 | 29,547 | +207% | 1 | 1 | 0% | 1,625 | 6,956 | +328% | 0 | 0 | — |
case-23 | pass→pass | 31,148 | 29,028 | -7% | 1 | 1 | 0% | 6,176 | 6,947 | +12% | 0 | 0 | — |
case-24 | pass→pass | 11,648 | 6,466 | -44% | 1 | 1 | 0% | 1,930 | 1,915 | -1% | 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. 24 cases were attempted. The headline lift of +17 percentage points is the difference between those two pass rates over the 24 comparable cases. 2 cases got worse with the skill loaded, and they are 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.