Install any skill in seconds. Free to start, no credit card required.
Get Started Free →SVG 曲线连接 + 便利贴节点 + 光标交互, 像白板 brainstorm
.claude/skills/nexu-io-frame-flowchart-sticky/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-19 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-21 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-04 | ✓→✗ | ▼ Worse | 201% | 0% |
| case-05 | ✓→✗ | ▼ Worse | 11% | 0% |
| case-06 | ✓→✗ | ▼ Worse | 11% | 0% |
【模板: 便利贴流程图帧 (Sticky Flowchart)】 【意图】把一个流程 / 系统 / 工作流画成"白板 + 便利贴"的样子, 适合 onboarding 视频、运营流程说明、系统架构讲解。Inspired by hyperframes flowchart。
【画布】1920×1080。背景: 米黄白板纸 #f4ede1 或冷灰白板 #f0f2f4; 加非常浅的 hex grid rgba(0,0,0,0.04) 让它有白板感。
【节点 (Sticky Notes)】
#fcd34d / 桃 #fca5a5 / 薄荷 #a7f3d0 / 天 #a5b4fc。transform: rotate(±2deg) 不一致, 投影 drop-shadow(0 6px 14px rgba(0,0,0,0.12)), 顶部胶带 linear-gradient(...) 装饰。Kalam / Caveat / Patrick Hand 手写感字体 (中文用 霞鹜文楷 或 LXGW WenKai Screen)。【连接线 (SVG)】
<path> Bezier 曲线连接节点, stroke #2a2a2a, width 2.5, stroke-linecap: round, stroke-dasharray: 0 (实线) 或 8 6 (虚线 = 条件分支)。marker-end, 黑色三角小箭头。【可选交互】
<svg> arrow + name tag), 浮在某节点旁, 模拟 figma 协作光标。【设计细节】
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-08 | fail→fail | 31,692 | 30,629 | -3% | 1 | 1 | 0% | 6,184 | 6,857 | +11% | 0 | 0 | — |
case-01 | fail→fail | 29,595 | 28,808 | -3% | 1 | 1 | 0% | 6,233 | 6,905 | +11% | 0 | 0 | — |
case-02 | fail→fail | 29,795 | 28,436 | -5% | 1 | 1 | 0% | 6,219 | 6,892 | +11% | 0 | 0 | — |
case-03 | fail→fail | 31,502 | 31,472 | -0% | 1 | 1 | 0% | 6,212 | 6,885 | +11% | 0 | 0 | — |
case-04 | pass→fail | 13,922 | 32,119 | +131% | 1 | 1 | 0% | 2,276 | 6,856 | +201% | 0 | 0 | — |
case-05 | pass→fail | 29,819 | 30,901 | +4% | 1 | 1 | 0% | 6,183 | 6,857 | +11% | 0 | 0 | — |
case-06 | pass→fail | 29,041 | 29,021 | -0% | 1 | 1 | 0% | 6,176 | 6,849 | +11% | 0 | 0 | — |
case-07 | fail→fail | 29,688 | 28,647 | -4% | 1 | 1 | 0% | 6,195 | 6,868 | +11% | 0 | 0 | — |
case-09 | fail→fail | 27,781 | 30,795 | +11% | 1 | 1 | 0% | 5,948 | 6,858 | +15% | 0 | 0 | — |
case-10 | fail→fail | 28,437 | 51,520 | +81% | 1 | 1 | 0% | 6,179 | 6,852 | +11% | 0 | 0 | — |
case-11 | fail→fail | 30,414 | 28,626 | -6% | 1 | 1 | 0% | 6,179 | 6,852 | +11% | 0 | 0 | — |
case-12 | fail→fail | 29,569 | 30,747 | +4% | 1 | 1 | 0% | 6,153 | 6,856 | +11% | 0 | 0 | — |
case-13 | fail→fail | 27,487 | 28,936 | +5% | 1 | 1 | 0% | 6,185 | 6,858 | +11% | 0 | 0 | — |
case-14 | fail→fail | 25,297 | 28,360 | +12% | 1 | 1 | 0% | 6,176 | 6,849 | +11% | 0 | 0 | — |
case-15 | fail→fail | 22,764 | 26,896 | +18% | 1 | 1 | 0% | 5,889 | 6,846 | +16% | 0 | 0 | — |
case-16 | fail→fail | 27,475 | 27,030 | -2% | 1 | 1 | 0% | 6,182 | 6,855 | +11% | 0 | 0 | — |
case-17 | fail→fail | 25,033 | 27,717 | +11% | 1 | 1 | 0% | 6,174 | 6,847 | +11% | 0 | 0 | — |
case-18 | fail→fail | 25,917 | 26,127 | +1% | 1 | 1 | 0% | 6,176 | 6,849 | +11% | 0 | 0 | — |
case-19 | fail→pass | 23,189 | 24,799 | +7% | 1 | 1 | 0% | 6,073 | 6,850 | +13% | 0 | 0 | — |
case-20 | fail→fail | 51,057 | 28,174 | -45% | 1 | 1 | 0% | 6,178 | 6,851 | +11% | 0 | 0 | — |
case-21 | fail→pass | 26,353 | 27,554 | +5% | 1 | 1 | 0% | 6,176 | 6,849 | +11% | 0 | 0 | — |
case-22 | fail→fail | 20,491 | 26,469 | +29% | 1 | 1 | 0% | 4,706 | 6,847 | +45% | 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. 3 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.