Slides As CodeSAFE

Intelligent Skill routing and workflow orchestration for AI agents — +21.12 pp reward, −29.6% tokens on SkillsBench with DeepSeekV4Flash-VE.

The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.

---
name: slides-as-code
description: "Build research slides with text-first source (Slidev/Marp/Reveal/Quarto) and reproducible export (PDF). Includes structure, figure reuse rules, and quality checklist for top-tier scientific presentations."
allowed-tools: [Read, Write, Edit, Bash, Grep, Glob]
---

# Slides as Code(顶级科研汇报的可复现幻灯片)

## 目标

解决常见问题:
- PPT 版本失控(final_v7)
- 图表重复截图、清晰度崩坏
- 汇报结构不稳定(信息密度失衡)

本 skill 让 slides 拥有:
- 文本源(Markdown / Quarto / LaTeX Beamer)
- 可复现导出(PDF、可选 PPTX)
- 图表复用(直接引用 `figures/` 最终导出,不截图)

---

## 技术路线(按约束选)

1) **Slidev**:开发者友好,组件化,适合组会/技术分享  
2) **Marp**:排版可控,适合快速生成简洁风格  
3) **Reveal.js**:网页交互强,适合 demo/交互展示  
4) **Quarto slides**:报告/论文/幻灯片一体化(推荐用于报告+slides联动)  

> 内容结构建议使用 `scientific-slides`;本 skill 更偏“工程化与导出管线”。

---

## 元规则(幻灯片质量的 80/20)

### 1) 一页一结论

- 标题就是结论,不是主题
- 每页只服务一个问题:What / Why / How / So what

### 2) 图不截图

- 图只从 `figures/**/out/` 引用(PDF/SVG/PNG)
- 禁止从论文 PDF 截图再贴回 PPT

### 3) 讲述顺序固定

推荐骨架:
1) 背景与缺口(1–2 页)
2) 方法直觉(1–2 页示意图)
3) 核心结果(3–6 页,每页一图一结论)
4) 局限与下一步(1–2 页)

---

## 模板

参考:`templates/slidev-structure.md`

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

What the source does

Filesystem

none-observed

Network

none-observed

Shell

none-observed

Dependencies

pinned

Secrets in source

none-found

Findings (0)

No findings outside the package's declared scope.

Gates applied: no_behavioural_pass.