According to AlphaMoat, as of 2026-09, PPT Presenter — 带逐字稿的演讲级PPT生成器 has 4,317 downloads and 2 stars — ranked #2,282 of 5,078 AI skills overall, and #194 of 435 in Office.
| Platform | Installs | Downloads | Stars | Comments |
|---|---|---|---|---|
| ClawHub ↗ | 31 | 1.2K | - | - |
| SkillHub ↗ | 980 | 4.3K | 2 | — |
Platforms count differently and update at different times.
Starting from a topic, notes, or markdown files, it produces an outline with page count, then per-slide titles, board descriptions, and speaker notes, plus one generated image per slide. The result is an HTML deck with reveal.js loaded from a CDN and no local dependencies, and image generation needs a Gemini API key. The deck loads reveal.js from a CDN and sends image-generation requests to Gemini; avoid confidential content unless approved for those services.
Profile published 2026-10-27
Generated from the skill’s documentation as examples; not results of actual runs.
The counting method changed in 2026-09: skills with the same name are now counted separately by author, and the full catalog is covered. Dashed lines show months before the change; they are not compared with later months, and no growth is computed across the change.
| Month | Downloads | MoM | Stars |
|---|---|---|---|
| 2026-09 | 4.3K | — | 2 |
| 2026-08 | 3.7K | +4.4% | 2 |
| 2026-07 | 3.6K | — | 2 |
| 2026-05 | 2.9K | — | 1 |
| 2026-03 | 1.5K | — | - |
The author’s original file, shown as plain text without edits or review by AlphaMoat. Links are not clickable; check commands before running them. Fetched 2026-10-23.
--- name: ppt-presenter description: Create professional HTML presentations with complete word-for-word speaker scripts (逐字稿) and presenter view. Use when user asks to create a presentation, PPT, slides, tech sharing, keynote, or talk. Core feature: every slide comes with a full speaker script (150-300 words) visible in presenter view (press S) — so you can present confidently without memorizing anything. Full pipeline: (1) content planning and outline, (2) per-slide content with three-audience-layer explanations, (3) AI-generated whiteboard illustrations via Gemini API, (4) reveal.js HTML assembly with lightbox, (5) word-for-word 逐字稿 in presenter view. Also use when asked to add speaker notes, generate slide images, or convert markdown notes into a presentation. --- # PPT Presenter — 带逐字稿的演讲级PPT生成器 为每一页生成完整演讲逐字稿 + 演讲者视图,让你自信上台不怯场。 ## Pipeline Overview
```
User Topic/Notes
↓
1. Content Planning (outline + page count)
↓
2. Per-Slide Content (title, board description, speaker notes)
↓
3. Image Generation (Gemini 3 Pro Image, one per slide)
↓
4. HTML Assembly (reveal.js + lightbox + presenter view)
↓
5. Speaker Scripts (word-for-word 逐字稿 in <aside class="notes">)
```
## Step 1: Content Planning
From user's topic, notes, or markdown files, produce an outline:
- Determine audience (technical level, mixed audiences)
- Plan 15-25 slides grouped into 4-7 sections
- Each section gets a divider slide (colored section number + title)
- Structure: Opening hook → Sections → Quick Start → Takeaway + Q&A
If user provides existing markdown/text content, read it and restructure into slide outline. Ask user to confirm page count and structure before proceeding.
## Step 2: Per-Slide Content
For each slide, define:
- **Page title** — concise, with emoji prefix
- **Visual description** — what to draw on a whiteboard (colors, layout, diagrams)
- **Key content** — bullet points, tables, code blocks, comparison charts
- **Speaker script** — 150-300 words per slide, conversational tone, can be read aloud directly
For mixed audiences, write three-layer explanations:
- 编程小白 (beginners): metaphors, daily-life analogies
- 职场白领 (business): ROI, efficiency, practical scenarios
- 开发人员 (developers): architecture, code, protocols
Write content to a markdown file first for user review, then proceed to image generation.
## Step 3: Image Generation
Generate one whiteboard-style illustration per slide using Gemini API.
Use the script: `scripts/generate_slide_images.py`
```bash
python3 scripts/generate_slide_images.py \
--prompts-file prompts.json \
--output-dir ./images \
--api-key "$GEMINI_API_KEY"
```
### Prompt Guidelines
- Style: "whiteboard marker drawing, hand-drawn educational style, colorful markers (red, blue, green, orange)"
- Include Chinese text in prompts when audience is Chinese
- Describe specific visual elements: diagrams, flowcharts, comparison tables, icons
- Keep prompts under 500 chars for best results
### API Details
- Model: `gemini-3-pro-image-preview` (preferred) or `imagen-4.0-generate-001`
- Endpoint: `https://generativelanguage.googleapis.com/v1beta/models/{model}:generateContent`
- Request body for Gemini 3:
```json
{
"contents": [{"parts": [{"text": "Generate this image: {prompt}"}]}],
"generationConfig": {"responseModalities": ["IMAGE", "TEXT"]}
}
```
- Response: `candidates[0].content.parts[].inlineData.data` (base64)
- Rate limit: add 2-3 second delay between requests
- Retry on network errors (SSL EOF, connection reset)
### Finding API Key
Check TOOLS.md for `Gemini Image Generation` section, or ask user for API key.
## Step 4: HTML Assembly
Use the template: `assets/reveal-template.html`
### Structure
```html
<!DOCTYPE html>
<html lang="zh-CN">
<head>
<!-- reveal.js 5.1.0 from CDN -->
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/reveal.css">
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/[email protected]/dist/theme/black.css">
<!-- Custom styles (see template) -->
</head>
<body>
<div class="reveal"><div class="slides">
<!-- Section dividers: colored background gradient + section number -->
<section data-background-gradient="...">
<div class="section-num">1</div>
<h2>Section Title</h2>
</section>
<!-- Content slides -->
<section>
<h2>Slide Title</h2>
<!-- Content: .two-col, .three-col, .card, table, .code-block, .timeline -->
<aside class="notes">Speaker script here...</aside>
</section>
</div></div>
<!-- Lightbox overlay (see template) -->
<!-- reveal.js init + lightbox JS (see template) -->
</body>
</html>
```
### Key CSS Classes (defined in template)
| Class | Use |
|-------|-----|
| `.card` | Dark rounded content box |
| `.two-col` / `.three-col` | Flex column layouts |
| `.code-block` | Styled code with `.comment`, `.cmd`, `.flag` |
| `.warning-box` | Red-border alert box |
| `.slide-img` | Clickable image (max-height 420px) |
| `.flow-box` + `.flow-arrow` | Horizontal flow diagrams |
| `.timeline` + `.timeline-item` | Vertical timeline with colored dots |
| `.tag-blue/green/red/orange/purple` | Colored pill tags |
| `.accent/.green/.red/.blue/.purple/.yellow/.pink` | Text colors |
| `.checklist` | Checkbox-style list |
| `.photo-row` | Horizontal image gallery |
| `.gradient-text` | Gradient colored heading |
### Lightbox Feature
All images with `.slide-img`, `.slide-img-large`, `.slide-img-full`, or inside `.photo-row` are clickable. Clicking opens fullscreen lightbox with:
- Left/right navigation (arrows + keyboard)
- Counter (current/total)
- Close button + ESC key
- Pauses reveal.js keyboard while open
The lightbox JS is included in the template.
### reveal.js Configuration
```javascript
Reveal.initialize({
hash: true,
slideNumber: 'c/t',
transition: 'slide',
width: 1280,
height: 720,
margin: 0.06,
plugins: [RevealHighlight, RevealNotes]
});
```
Keyboard hints shown at bottom: `← → 翻页 · S 演讲者视图 · F 全屏 · O 总览`
## Step 5: Speaker Scripts
Add `<aside class="notes">` to every content slide (not section dividers).
### Script Guidelines
- **Length**: 150-300 words per slide (1-2 minutes speaking time)
- **Tone**: Conversational, as if speaking to audience directly
- **Structure**: Hook → explain visual → key points → transition to next
- **Language**: Match audience language (Chinese for Chinese audience)
- **Include**: Analogies, real examples, audience engagement ("大家看...")
- **Avoid**: Reading bullet points verbatim; scripts should expand on slide content
### Presenter View
User presses `S` to open presenter view showing:
- Current slide (left)
- Speaker notes/script (right)
- Next slide preview
- Timer
## Output Checklist
Before delivering, verify:
- [ ] All slides have titles with emoji prefixes
- [ ] All content slides have `<aside class="notes">` with full scripts
- [ ] All slide images generated and saved to `images/` directory
- [ ] Images use `.slide-img` class (clickable lightbox)
- [ ] Section dividers have colored backgrounds and section numbers
- [ ] Lightbox JS is included and functional
- [ ] reveal.js loads from CDN (no local dependencies)
- [ ] File opens correctly in browser (`open index.html`)
A full manual for creating, editing, and managing Tencent Docs through MCP.
Creates, inspects, and edits Excel files while keeping formulas, formatting, and workbook structure intact.
A working guide for Word and DOCX documents that keeps styles, numbering, and layout intact after edits.
Access Gmail, Calendar, Drive, Contacts, Sheets and Docs from the command line.
PDF and Image Text Extractor organizes document text, tables and scanned-page recognition into text results.
A skill that generates PPT decks from a topic and style.
The official skill for working with Kdocs cloud documents through kdocs-cli.
A small CLI that edits a chosen PDF page from a natural language instruction.
Data month: 2026-09 · Downloads, stars and installs are aggregated monthly from public skill registries (ClawHub, SkillHub). See methodology.