Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →PPT 演示助手 - 幻灯片设计、内容结构、视觉优化、动画建议、演讲稿
.claude/skills/dongsheng123132-ppt-designer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-19 | ✗→✓ | ▲ Improved | 51% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 0% | 0% |
专业的演示文稿设计工具，帮你打造高效的商务演讲。
1. 封面：标题 + 演讲者 + 日期
2. 目录：内容概览
3. 背景：为什么要做这个项目
4. 目标：我们要达成什么
5. 方法：怎么做
6. 进展：目前做到了什么程度
7. 数据：关键成果展示
8. 分析：数据说明了什么
9. 计划：下一步做什么
10. 结束：谢谢 + Q&A1. 封面：产品名称 + 品牌
2. 目录
3. 市场需求：痛点分析
4. 产品介绍：核心功能
5. 技术架构：创新点
6. 产品演示：截图/视频
7. 竞品对比：优势凸显
8. 用户案例：成功故事
9. 市场前景：规模预测
10. 商业模式：盈利方式
11. 推广策略：渠道规划
12. 团队介绍：核心成员
13. 里程碑：发展历程
14. 融资计划：资金用途
15. 结束页：联系方式| 页面 | 文字量 | 重点 | |------|--------|------| | 封面 | 标题 + 副标题 | 简洁有力 | | 目录 | 3-5 个章节 | 结构清晰 | | 内容页 | 6x6 法则 | 一页一观点 | | 图表页 | 标题 + 图表 + 结论 | 数据说话 | | 结束页 | 谢谢 + 联系方式 | 行动号召 |
| 场景 | 主色 | 辅色 | |------|------|------| | 商务汇报 | 蓝色 #2C5282 | 灰色 #718096 | | 科技产品 | 深蓝 #1A365D | 青色 #319795 | | 金融投资 | 金色 #D69E2E | 深灰 #2D3748 | | 医疗健康 | 绿色 #276749 | 浅绿 #68D391 | | 教育培训 | 橙色 #DD6B20 | 暖灰 #F7FAFC |
帮我设计一个产品发布会 PPT，包含 15 页这是一份技术方案，帮我转成演示文稿结构这张幻灯片内容太拥挤了，帮我精简提供一个互联网产品路演的 PPT 结构| 功能 | 快捷键 | |------|--------| | 新建幻灯片 | Ctrl+M | | 复制幻灯片 | Ctrl+D | | 放映 | F5 | | 画笔标注 | Ctrl+P | | 激光笔 | Ctrl+L |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-19 | fail→pass | 13,903 | 11,853 | -15% | 1 | 1 | 0% | 1,835 | 2,770 | +51% | 0 | 0 | — |
case-01 | fail→fail | 26,645 | 27,824 | +4% | 1 | 1 | 0% | 3,918 | 4,756 | +21% | 0 | 0 | — |
case-02 | fail→fail | 25,096 | 18,252 | -27% | 1 | 1 | 0% | 3,815 | 4,040 | +6% | 0 | 0 | — |
case-03 | fail→fail | 26,014 | 27,333 | +5% | 1 | 1 | 0% | 3,342 | 5,380 | +61% | 0 | 0 | — |
case-04 | fail→fail | 22,185 | 14,636 | -34% | 1 | 1 | 0% | 2,976 | 3,286 | +10% | 0 | 0 | — |
case-05 | fail→fail | 17,174 | 22,528 | +31% | 1 | 1 | 0% | 2,715 | 4,234 | +56% | 0 | 0 | — |
case-06 | fail→pass | 19,371 | 18,646 | -4% | 1 | 1 | 0% | 3,117 | 3,591 | +15% | 0 | 0 | — |
case-07 | pass→pass | 5,536 | 3,553 | -36% | 1 | 1 | 0% | 1,040 | 1,556 | +50% | 0 | 0 | — |
case-08 | pass→pass | 10,351 | 6,331 | -39% | 1 | 1 | 0% | 1,832 | 1,845 | +1% | 0 | 0 | — |
case-09 | fail→pass | 14,300 | 12,963 | -9% | 1 | 1 | 0% | 2,200 | 2,681 | +22% | 0 | 0 | — |
case-10 | fail→pass | 15,275 | 12,825 | -16% | 1 | 1 | 0% | 2,342 | 2,716 | +16% | 0 | 0 | — |
case-11 | fail→pass | 18,224 | 9,902 | -46% | 1 | 1 | 0% | 2,506 | 2,501 | -0% | 0 | 0 | — |
case-12 | pass→pass | 18,929 | 16,615 | -12% | 1 | 1 | 0% | 2,676 | 3,213 | +20% | 0 | 0 | — |
case-13 | pass→pass | 16,220 | 18,379 | +13% | 1 | 1 | 0% | 2,450 | 3,337 | +36% | 0 | 0 | — |
case-14 | fail→pass | 11,842 | 2,968 | -75% | 1 | 1 | 0% | 1,521 | 1,415 | -7% | 0 | 0 | — |
case-15 | pass→pass | 15,691 | 13,702 | -13% | 1 | 1 | 0% | 2,225 | 3,227 | +45% | 0 | 0 | — |
case-16 | fail→pass | 23,860 | 16,736 | -30% | 1 | 1 | 0% | 3,232 | 3,329 | +3% | 0 | 0 | — |
case-17 | fail→pass | 16,478 | 15,166 | -8% | 1 | 1 | 0% | 2,601 | 3,201 | +23% | 0 | 0 | — |
case-18 | fail→pass | 18,066 | 11,689 | -35% | 1 | 1 | 0% | 2,464 | 2,447 | -1% | 0 | 0 | — |
case-20 | pass→fail | 26,944 | 16,645 | -38% | 1 | 1 | 0% | 4,075 | 3,382 | -17% | 0 | 0 | — |
case-21 | pass→fail | 15,779 | 12,447 | -21% | 1 | 1 | 0% | 2,206 | 2,515 | +14% | 0 | 0 | — |
case-22 | fail→fail | 34,310 | 27,015 | -21% | 1 | 1 | 0% | 4,836 | 5,375 | +11% | 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 +32 percentage points is the difference between those two pass rates over the 22 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.