Figma Context MCP
Give your coding agent access to your Figma data. Implement designs in any framework in one-shot.
Speaker is a Codex skill project for academic presentations: read real.pptx, combine text extraction, PPTX structure parsing, page-by-page rendering, OCR, and visual review to generate page-by-page speaker notes, and write a clean version of the lecture into the PowerPoint comment area.
If you made this, paste this code into your README. The badge links back to this page.
[](https://mcp.popnx.com/en/skills/ai272-speaker)MCP servers that connect the AI to services for the same work
Give your coding agent access to your Figma data. Implement designs in any framework in one-shot.
draw.io diagrams your AI agent draws on a live browser canvas that you can also edit by hand
Create, inspect, validate, and save editable 3D building scenes with Pascal's MCP server.
Same kind of work, ordered like the Skills list
| This one speaker | impeccable | diagram-design | frontend-slides | guizang-ppt-skill | |
|---|---|---|---|---|---|
| Maker | AI272 | pbakaus | cathrynlavery | zarazhangrui | op7418 |
| Skills | 1 | 1 | 1 | 1 | – |
| GitHub stars | 479 | 110,690 | 67,371 | 32,358 | 29,835 |
| Last code change | Jun 23, 2026 | Nov 16, 2026 | Nov 13, 2026 | Jun 24, 2026 | Oct 30, 2026 |
| License | – | Apache-2.0 | MIT | MIT | AGPL-3.0 |
1 skills (instruction sets) by AI272 that an AI reads and follows · not an MCP server
MCP = the hands
Connects the AI to an outside service or data, such as reading Notion or creating an invoice. A program or server has to be running.
Skill = the manual
A text file that teaches the AI how to do a task. It connects to nothing.
Some overlap: a few Skills run scripts that call tools, and a few MCP servers only send knowledge. Start from the task you want done.
Claude CodeCodex
From the maker’s README and the folders in the repo
GitHub account AI272 (person), 2 years old
★ 479 stars, few users so far · code last changed Jun 23, 2026
6 scripts that can run on a computer: scripts/inject_notes.py, scripts/read_slides.py, scripts/render_slides.py, scripts/vision_review.py and 2 more
No internet calls found in the scripts
No hooks or writes to machine settings found
Checked automatically from the repo files at commit a7b5b9d (read only, nothing was run). Not a safety guarantee: read it and decide for yourself
First: which AI do you use?
Pick your app and this page shows only how to install it there
This skill has no license, so we do not hand out a ZIP from our site. Download it from GitHub and arrange the folder yourself, or use "Let the AI install it" above, which is easier
Download from GitHub (whole repo) ↗
Unzip it, then zip only the ppt-speech-writer folder again. The ZIP must have a ppt-speech-writer folder with SKILL.md at the top level, no loose files
+Create skillUpload a skillpick ppt-speech-writer.zip
SettingsCapabilitiesCode execution and file creation
Turn on Code execution and file creation first (newer menus: Create and edit files) · personal plans: Settings → Capabilities · Team/Enterprise: an admin turns it on under Organization settings → Plugins & skills · every plan, including Free
Upload once; it works in claude.ai and Claude Desktop, and in recent Claude Code signed in with a Claude account (not with an API key)
Cowork and cloud sessions use the skills turned on in your claude.ai account, not ~/.claude/skills on your computer (Claude Desktop scheduled tasks run locally and still load it)
Checked against official docs Nov 12, 2026 · Source: support.claude.com, support.claude.com, code.claude.com
Copy this sentence and paste it into a Claude Code chat
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Restart Claude Code, then type /ppt-speech-writer or ask for a matching task and Claude picks it up
Checked against official docs Nov 12, 2026 · Source: code.claude.com
A Skill is instructions the AI will follow. Read the files before installing, and install only from makers you trust.
Mac: press Command+Space and type Terminal · Windows: the command below does not work in PowerShell; use “Let AI install it” above instead
This downloads the repo to a temporary folder and copies only the folders that contain a SKILL.md
Checking the commands · copy buttons appear once they pass
The ChatGPT desktop app has Codex built in, and Codex reads skills from ~/.agents/skills
Open Codex in the ChatGPT desktop app and paste this sentence into the chat
The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Put the ppt-speech-writer folder at ~/.agents/skills/ppt-speech-writer
~ is your home folder ($HOME)
A Skill is instructions the AI will follow. Read the files before installing, and install only from makers you trust.
Mac: press Command+Space and type Terminal · Windows: the command below does not work in PowerShell; use “Let AI install it” above instead
This downloads the repo to a temporary folder and copies only the folders that contain a SKILL.md
Checking the commands · copy buttons appear once they pass
Checked against official docs Nov 15, 2026 · Source: learn.chatgpt.com
Skills you install yourself do not work here; only skills that come with a plugin from the directory
To use this skill with ChatGPT, use the ChatGPT desktop app as above
Checked against official docs Nov 15, 2026 · Source: learn.chatgpt.com
Copy this sentence and paste it into a Codex chat · Works in Codex inside the ChatGPT desktop app, Codex CLI and the IDE extension
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Codex picks up newly installed skills by itself; if one does not show up, restart Codex · To use it: in ChatGPT type @ and pick the skill · in Codex CLI or the IDE extension run /skills or type $ and the skill name
Checked against official docs Nov 15, 2026 · Source: learn.chatgpt.com
A Skill is instructions the AI will follow. Read the files before installing, and install only from makers you trust.
Mac: press Command+Space and type Terminal · Windows: the command below does not work in PowerShell; use “Let AI install it” above instead
This downloads the repo to a temporary folder and copies only the folders that contain a SKILL.md
Checking the commands · copy buttons appear once they pass
Copy this sentence and paste it into a Cursor chat
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
In Cursor’s Agent chat, type / and pick ppt-speech-writer
Checked against official docs Nov 12, 2026 · Source: cursor.com
A Skill is instructions the AI will follow. Read the files before installing, and install only from makers you trust.
Mac: press Command+Space and type Terminal · Windows: the command below does not work in PowerShell; use “Let AI install it” above instead
This downloads the repo to a temporary folder and copies only the folders that contain a SKILL.md
Checking the commands · copy buttons appear once they pass
For other apps that read skills. The sentence gives the paths for Claude Code, Codex and Cursor so the AI picks the one for your app · Copy this sentence into the app’s chat
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Checked against official docs Nov 12, 2026 · Source: code.claude.com, learn.chatgpt.com, cursor.com
Other apps that support Agent Skills (such as Gemini CLI or GitHub Copilot) use different folders. Check that app’s guide, then copy the folder with the SKILL.md there.
Skill folders checked against the official Claude Code, Codex and Cursor guides on Nov 11, 2026
The text below comes from the maker’s repo, not from PopNx MCP
Showing the first 8,000 characters · Open the full file on GitHub
---
name: ppt-speech-writer
description: Read a real .pptx using text extraction, OOXML inspection, slide rendering, OCR, visual inventory, and vision-capable screenshot review; then write academic speaker notes grounded in every visible slide element, generate a complete display-version document, and inject clean notes into the PowerPoint notes pane. Use when the user wants speaker notes, presenter notes, a speech script, narration, or annotated notes for an existing PowerPoint deck, especially when slides contain images, charts, tables, SmartArt, axes, legends, or screenshot text.
---
# PPT Speech Writer
You are a senior academic presentation coach. This skill writes slide-by-slide speaker notes for an existing `.pptx`, grounded in the actual visible deck. It must inspect both the structured PowerPoint content and rendered slide images before drafting.
## Grounding Contract
Do not rely on text boxes alone. A slide is considered read only after these evidence sources have been checked:
1. Structured extraction from PowerPoint objects: text frames, tables, chart XML, pictures, placeholders, notes, and raw OOXML text.
2. Rendered slide screenshots, one image per slide.
3. OCR or visual inspection of rendered slides when screenshots, charts, diagrams, SmartArt, or image-contained text are present.
4. A visible-element inventory for every slide.
5. Vision-capable review of rendered screenshots for every slide with charts, diagrams, SmartArt, screenshots, dense figures, or image-only content.
If a visible element cannot be interpreted reliably, say so and ask the user before writing notes for that slide. Never invent chart values, axes, labels, image meaning, or screenshot text.
## Language Lock
Do not infer the output language from the user's chat language. Before writing any notes, explicitly confirm exactly one output language:
- English
- Chinese
- same as the deck language
- another user-specified language
Never draft speaker notes, display notes, glossary entries, timing-table labels, transitions, coverage notes, or injected clean notes until the output language is confirmed.
Once confirmed, use that language consistently across the entire deliverable. Technical terms may remain in their canonical form, such as `PPO`, `AUROC`, `PowerPoint`, `SmartArt`, or dataset names, but sentence grammar, explanations, labels, table headers, and transitions must follow the selected language.
If the selected language is English:
- Write all prose, transitions, labels, glossary definitions, timing-table headers, and coverage notes in English.
- If a slide contains Chinese or Japanese text, quote only the necessary original term and immediately explain it in English.
- Do not write mixed sentences such as "This model 说明了 robustness."
If the selected language is Chinese:
- Write all prose, transitions, labels, glossary definitions, timing-table headers, and coverage notes in Chinese.
- Keep standard technical names in English only when they are the canonical term.
- Do not write mixed sentences such as "这个 model shows strong robustness."
- Embed English technical terms naturally in Chinese syntax, for example: "`AUROC` 用来衡量模型区分正负样本的能力。"
## Slide Prose Style
Do not begin slide notes by describing the slide object. Begin with the claim, implication, finding, method role, or argument step.
Banned English openings:
- "This slide shows..."
- "This slide presents..."
- "This slide explains..."
- "On this slide..."
- "Here we can see..."
- "The slide is about..."
Banned Chinese openings:
- "这一页展示了..."
- "这一页说明了..."
- "这一页主要讲..."
- "在这一页中..."
- "我们可以看到..."
- "这页是关于..."
Preferred pattern:
- Weak: "This slide shows the optimization setup."
- Strong: "The experiments use a fixed optimization protocol so later comparisons stay controlled."
- Weak: "这一页展示了实验设置。"
- Strong: "实验设置被固定下来,是为了保证后续结果比较具有可解释性。"
Write speaker notes as a coherent oral argument, not as captions for slides. Each page should open with a content-level thesis sentence, then explain the visible evidence that supports it.
## Timing And Pacing Model
Speaker notes must fit the spoken duration when read aloud at a realistic pace, with pauses. Do not treat raw word count as speaking time. Plan against pause-adjusted rates, then verify the total against the clock.
### Planning rates (pauses already absorbed)
These rates already account for natural pauses, breaths, and emphasis. Use them as the default budget:
| Output language | Planning rate | Notes |
|-----------------|---------------|-------|
| English | 110 words/min | Academic delivery. Range 100–130. Use 100 for dense or non-native delivery. |
| Chinese | 165 字/min | Academic delivery. Range 150–180. Count Chinese characters, not words. |
Calibration rule: if the speaker says a previous talk ran long, trust the measured pace over the default. Example: a 15-minute talk that actually took 30 minutes ran at `1800 ÷ 30 = 60` effective words per minute. When the speaker reports such a history, recompute with their real rate instead of the table default.
### Reserve time before budgeting words
Do not spend the whole clock on words. From the target duration, subtract:
- Slide transitions: about 3 seconds per slide.
- Deliberate `[PAUSE]` markers: about 1.5 seconds each.
- A global safety buffer of 10–15% for audience reactions, demos, and overruns.
Budget formula for a talk of `T` minutes across `N` slides:
```text
gross_seconds = T * 60
reserved = 3 * N + 1.5 * pause_count + 0.12 * gross_seconds
usable_seconds = gross_seconds - reserved
english_word_budget = usable_seconds / 60 * 110
chinese_char_budget = usable_seconds / 60 * 165
```
Worked example — 15-minute English talk, 12 slides, ~24 planned pauses:
```text
gross_seconds = 900
reserved = 36 (transitions) + 36 (pauses) + 108 (12% buffer) = 180
usable_seconds = 720
word_budget = 720 / 60 * 110 ≈ 1,320 words
```
So a 15-minute English talk targets roughly 1,300–1,400 spoken words, not 1,800. The earlier 1,800-word draft was about a third over budget, which is exactly why it overran. Apply the same arithmetic with the Chinese rate for a Chinese talk.
### Per-slide budget
Distribute the total budget by content weight, not evenly. A title or section divider may take 30–50 words; a dense results slide may take 150–180. Record each slide's target in the timing table `budget` field and the actual count in `word_count`, and the number of `[PAUSE]` markers in `pauses`. If a slide's `word_count` exceeds its `budget`, compress the prose or recommend splitting the slide. Never let total `word_count` exceed the computed budget.
### Writing to the budget in both languages
- Keep spoken sentences short, under 20 words. 中文每句也尽量短,便于换气。
- Place `[PAUSE]` after a thesis sentence or before a key number, not at random.
- Do not pad to hit a budget. A slightly short talk is safer than an overrun.
- Compute the Chinese and English drafts against their own language rate. Do not reuse the English word count as the Chinese character count; the same idea is usually fewer characters in Chinese than words in English.
## Required Workflow
### 1. Create Output Layout
Keep user-facing deliverables separate from intermediate evidence files.
Use this layout:
```text
<deck-stem>-speaker-output/
├── <deck-stem>-with-notes.pptx
├── <deck-stem>-display.docx
├── <deck-stem>-display.md # only if python-docx is unavailable
├── <deck-stem>-vision-review.md # optional, only if the Markdown review is requested
└── work/
├── slide_extract.json
├── visual_inventory.json
├── vision_review_packet.json
├── vision_review.json
├── display_document.json
├── notes.json
└── rendered_slides/
```
In the final response, surface only the user-facing deliverables as a short summary plus their file paths:
- PowerPoint with speaker notes
- complete display rehearsal document
- vision-review packet, and the Markdown version only if it was generated
All Ask, share a tip or report that it does not work. Sorted by what people marked helpful.
No posts yet. Be the first to ask or share.
Posts here are written by users, not the PopNx team.