Install any skill in seconds. Free to start, no credit card required.
Get Started Free →电子杂志 × 电子墨水; 10 个版面 + 5 套调色板 (墨水/靛蓝瓷/森林墨/牛皮纸/沙丘)
.claude/skills/nexu-io-deck-guizang-editorial/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-20 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 16% | 0% |
【模板: 归藏编辑墨水 Deck (Editorial × E-Ink)】 【意图】叙事、观点、分享、个人风格表达。墨纸印刷感, 不要科技感。Inspired by op7418/guizang-ppt-skill Style A。
【调色板 — 5 选 1, 严禁改 hex、严禁混用】
#0a0a0b, paper #f1efea, paper-tint #e8e5de, ink-tint #18181a. 默认 / 通用商业 / 科技。#0a1f3d, paper #f1f3f5, paper-tint #e4e8ec, ink-tint #152a4a. 科技 / 研究 / 数据。#1a2e1f, paper #f5f1e8, paper-tint #ece7da, ink-tint #253d2c. 自然 / 可持续 / 文化。#2a1e13, paper #eedfc7, paper-tint #e0d0b6, ink-tint #3a2a1d. 怀旧 / 人文 / 文学。#1f1a14, paper #f0e6d2, paper-tint #e3d7bf, ink-tint #2d2620. 艺术 / 设计 / 时尚。【布局 — 10 个磁带式版式池, 可复用; 数量由【用户内容】决定, 完整覆盖每个要点; 短内容 6-12 张起步, 长内容应更多 (同一版式可在不同章节重复使用)】
【设计细节】
Playfair Display (英) / Noto Serif SC (中); Body 用 Inter / Noto Sans SC; 编号 / 数字偶尔可用 italic 衬线。01 / 12; 顶部细 hairline rule + 期刊 logo / topic。| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→fail | 33,304 | 31,624 | -5% | 1 | 1 | 0% | 6,218 | 7,189 | +16% | 0 | 0 | — |
case-20 | fail→pass | 30,875 | 29,708 | -4% | 1 | 1 | 0% | 6,169 | 7,140 | +16% | 0 | 0 | — |
case-02 | fail→pass | 32,126 | 28,778 | -10% | 1 | 1 | 0% | 6,208 | 7,179 | +16% | 0 | 0 | — |
case-03 | pass→pass | 30,530 | 31,935 | +5% | 1 | 1 | 0% | 6,205 | 7,177 | +16% | 0 | 0 | — |
case-04 | fail→pass | 29,897 | 28,400 | -5% | 1 | 1 | 0% | 6,190 | 7,161 | +16% | 0 | 0 | — |
case-05 | fail→pass | 31,193 | 31,127 | -0% | 1 | 1 | 0% | 6,182 | 7,153 | +16% | 0 | 0 | — |
case-06 | fail→pass | 31,797 | 28,626 | -10% | 1 | 1 | 0% | 6,174 | 7,145 | +16% | 0 | 0 | — |
case-07 | fail→pass | 31,471 | 29,644 | -6% | 1 | 1 | 0% | 6,173 | 7,144 | +16% | 0 | 0 | — |
case-08 | fail→fail | 30,728 | 28,571 | -7% | 1 | 1 | 0% | 6,178 | 7,149 | +16% | 0 | 0 | — |
case-09 | fail→fail | 30,866 | 30,184 | -2% | 1 | 1 | 0% | 6,175 | 7,146 | +16% | 0 | 0 | — |
case-10 | fail→pass | 28,319 | 29,762 | +5% | 1 | 1 | 0% | 6,173 | 7,144 | +16% | 0 | 0 | — |
case-11 | pass→pass | 30,904 | 31,987 | +4% | 1 | 1 | 0% | 6,177 | 7,148 | +16% | 0 | 0 | — |
case-21 | pass→fail | 28,457 | 29,419 | +3% | 1 | 1 | 0% | 6,222 | 7,193 | +16% | 0 | 0 | — |
case-12 | pass→pass | 28,209 | 29,595 | +5% | 1 | 1 | 0% | 6,179 | 7,150 | +16% | 0 | 0 | — |
case-13 | pass→fail | 29,236 | 29,484 | +1% | 1 | 1 | 0% | 6,057 | 7,145 | +18% | 0 | 0 | — |
case-14 | fail→fail | 18,269 | 29,703 | +63% | 1 | 1 | 0% | 3,736 | 7,152 | +91% | 0 | 0 | — |
case-15 | pass→pass | 22,979 | 29,280 | +27% | 1 | 1 | 0% | 4,635 | 7,139 | +54% | 0 | 0 | — |
case-22 | pass→fail | 28,714 | 30,347 | +6% | 1 | 1 | 0% | 6,200 | 7,171 | +16% | 0 | 0 | — |
case-16 | fail→fail | 31,187 | 31,527 | +1% | 1 | 1 | 0% | 6,174 | 7,145 | +16% | 0 | 0 | — |
case-17 | fail→fail | 30,997 | 29,186 | -6% | 1 | 1 | 0% | 6,171 | 7,142 | +16% | 0 | 0 | — |
case-18 | fail→fail | 30,467 | 28,664 | -6% | 1 | 1 | 0% | 6,172 | 7,143 | +16% | 0 | 0 | — |
case-19 | pass→pass | 19,519 | 28,731 | +47% | 1 | 1 | 0% | 4,047 | 7,147 | +77% | 0 | 0 | — |
case-23 | pass→fail | 30,860 | 29,686 | -4% | 1 | 1 | 0% | 6,205 | 7,176 | +16% | 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. 5 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/12/2026 | +9% |
Other measured skills in the registry, with their headline benchmark lift.