Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Word 文档助手 - 文章撰写、格式排版、模板选择、目录生成、批量处理
.claude/skills/dongsheng123132-word-writer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 38% | 0% |
| case-22 | ✗→✓ | ▲ Improved | -42% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -39% | 0% |
| case-20 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-21 | ✗→✓ | ▲ Improved | -41% | 0% |
专业的 Word 文档创作和排版工具，帮你快速生成规范的商务文档。
【封面】标题 + 副标题 + 日期 + 公司名
【目录】自动生成
【前言】背景、目的、范围
【正文】分章节，每章含图表
【结论】核心发现 + 建议
【附录】数据来源、参考资料1. 背景分析（SWOT）
2. 目标设定（SMART）
3. 执行计划（时间表 + 责任人）
4. 资源需求（预算 + 人员）
5. 风险预案| 功能 | 快捷键 | |------|--------| | 加粗 | Ctrl+B | | 斜体 | Ctrl+I | | 下划线 | Ctrl+U | | 标题1 | Ctrl+Alt+1 | | 标题2 | Ctrl+Alt+2 | | 插入目录 | Ctrl+Shift+O |
帮我写一份项目立项报告，主题是"电商平台用户增长方案"这篇论文的格式太乱了，帮我调整成学术规范格式生成一个产品介绍文档的模板，包含产品特点、优势、应用场景| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 15,919 | 24,358 | +53% | 1 | 1 | 0% | 2,310 | 3,190 | +38% | 0 | 0 | — |
case-02 | pass→pass | 10,405 | 8,122 | -22% | 1 | 1 | 0% | 1,683 | 1,514 | -10% | 0 | 0 | — |
case-03 | pass→pass | 12,884 | 8,202 | -36% | 1 | 1 | 0% | 1,642 | 1,745 | +6% | 0 | 0 | — |
case-04 | pass→pass | 15,598 | 17,379 | +11% | 1 | 1 | 0% | 2,370 | 2,922 | +23% | 0 | 0 | — |
case-05 | pass→pass | 7,260 | 9,228 | +27% | 1 | 1 | 0% | 1,139 | 1,632 | +43% | 0 | 0 | — |
case-06 | pass→pass | 5,927 | 4,824 | -19% | 1 | 1 | 0% | 994 | 1,318 | +33% | 0 | 0 | — |
case-07 | pass→pass | 12,352 | 8,434 | -32% | 1 | 1 | 0% | 1,757 | 1,566 | -11% | 0 | 0 | — |
case-22 | fail→pass | 18,992 | 8,343 | -56% | 1 | 1 | 0% | 2,904 | 1,678 | -42% | 0 | 0 | — |
case-08 | pass→pass | 7,221 | 5,877 | -19% | 1 | 1 | 0% | 1,118 | 1,253 | +12% | 0 | 0 | — |
case-09 | fail→pass | 9,885 | 3,367 | -66% | 1 | 1 | 0% | 1,649 | 1,001 | -39% | 0 | 0 | — |
case-10 | pass→pass | 3,935 | 2,561 | -35% | 1 | 1 | 0% | 702 | 1,014 | +44% | 0 | 0 | — |
case-11 | pass→pass | 3,367 | 2,819 | -16% | 1 | 1 | 0% | 631 | 985 | +56% | 0 | 0 | — |
case-12 | pass→pass | 12,806 | 13,209 | +3% | 1 | 1 | 0% | 2,069 | 2,544 | +23% | 0 | 0 | — |
case-13 | pass→pass | 14,859 | 12,741 | -14% | 1 | 1 | 0% | 2,230 | 2,610 | +17% | 0 | 0 | — |
case-14 | pass→pass | 7,546 | 14,324 | +90% | 1 | 1 | 0% | 1,024 | 2,363 | +131% | 0 | 0 | — |
case-15 | pass→pass | 12,573 | 7,798 | -38% | 1 | 1 | 0% | 1,637 | 1,779 | +9% | 0 | 0 | — |
case-16 | pass→pass | 13,176 | 9,530 | -28% | 1 | 1 | 0% | 1,777 | 2,041 | +15% | 0 | 0 | — |
case-17 | pass→pass | 17,624 | 15,136 | -14% | 1 | 1 | 0% | 2,282 | 2,711 | +19% | 0 | 0 | — |
case-18 | pass→pass | 12,400 | 11,804 | -5% | 1 | 1 | 0% | 2,171 | 2,564 | +18% | 0 | 0 | — |
case-19 | pass→pass | 7,764 | 6,619 | -15% | 1 | 1 | 0% | 1,060 | 1,546 | +46% | 0 | 0 | — |
case-20 | fail→pass | 16,732 | 6,856 | -59% | 1 | 1 | 0% | 2,486 | 1,622 | -35% | 0 | 0 | — |
case-21 | fail→pass | 19,069 | 8,303 | -56% | 1 | 1 | 0% | 2,854 | 1,684 | -41% | 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 +23 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.