REA
Reverse engineer anything from your terminal or agent with one CLI and MCP server.
Talks as beats, not slides: a fixed 1920×1080 React stage where one click is one beat. Offline, deterministic, presenter view, and a Claude Code skill.
如果你是作者,可把这段代码贴到 README。徽章会链接回本页。
[](https://mcp.popnx.com/en/skills/borjaperfra-beatdeck)在同一工作中把 AI 连到服务的 MCP
Reverse engineer anything from your terminal or agent with one CLI and MCP server.
Codebase intelligence for TypeScript and JavaScript, exposed as structured MCP analysis tools.
MCP server for terminal commands, file operations, and process management
同类工作,排序与 Skill 列表相同
| 本项 beatdeck | rea | andrej-karpathy-skills | graphify | MoneyPrinterTurbo | |
|---|---|---|---|---|---|
| 作者 | borjaperfra | morluto | multica-ai | Graphify-Labs | harry0703 |
| Skill 数 | 1 | 1 | 1 | 1 | 1 |
| GitHub 星标 | 244 | 465,460 | 233,910 | 149,364 | 144,149 |
| 最近代码更新 | 2026年10月2日 | 2026年11月17日 | 2026年4月20日 | 2026年11月17日 | 2026年11月17日 |
| 许可证 | – | MIT | – | Apache-2.0 | MIT |
borjaperfra 提供的 1 个 Skill(指令集),AI 读取后照做 · 不是 MCP server
MCP = 手
把 AI 连到外部服务或数据,例如读取 Notion 或开发票。需要有程序或服务器在运行。
Skill = 手册
教 AI 怎么做某项工作的文本文件,不连接任何服务。
有些会重叠:少数 Skill 带有调用工具的脚本,少数 MCP 只提供知识。请从你想完成的工作出发。
Claude Code
依据开发者 README 和 repo 中的文件夹
GitHub account borjaperfra (person), 3 years old
★ 244 stars, few users so far · code last changed 2026年10月2日
1 scripts that can run on a computer: scripts/extract_pptx.py
No internet calls found in the scripts
No hooks or writes to machine settings found
Checked automatically from the repo files at commit 329a971 (read only, nothing was run). Not a safety guarantee: read it and decide for yourself
第一步:你用哪个 AI?
选择你用的应用,本页只显示该应用的安装方法
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 skills/building-a-beatdeck folder again. The ZIP must have a building-a-beatdeck folder with SKILL.md at the top level, no loose files
+Create skillUpload a skillpick building-a-beatdeck.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 2026年11月12日 · Source: support.claude.com, support.claude.com, code.claude.com
复制这句话,粘贴到 Claude Code 的对话中
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
重新打开 Claude Code,输入 /building-a-beatdeck,或直接提出相关任务,Claude 会自动选用
Checked against official docs 2026年11月12日 · Source: code.claude.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
正在检查命令 · 通过后才会显示复制按钮
摘自开发者 README,会下载并运行开发者的代码
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
ChatGPT 桌面应用内置 Codex,Codex 从 ~/.agents/skills 文件夹读取 Skill
在 ChatGPT 桌面应用中打开 Codex,把这句话粘贴到对话中
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
把 building-a-beatdeck 文件夹放到 ~/.agents/skills/building-a-beatdeck
~ 是你的主文件夹($HOME)
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
不能使用你自己安装的 Skill,只能使用目录中插件自带的 Skill
想在 ChatGPT 中使用这个 Skill,请按上面的方法使用 ChatGPT 桌面应用
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
复制这句话,粘贴到 Codex 的对话中 · 适用于 ChatGPT 桌面应用中的 Codex、Codex CLI 和 IDE 扩展
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 会自动发现新安装的 Skill;如果没有出现,请重启 Codex · 使用方法:在 ChatGPT 中输入 @ 并选择 Skill · 在 Codex CLI 或 IDE 扩展中运行 /skills,或输入 $ 加 Skill 名称
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
复制这句话,粘贴到 Cursor 的对话中
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
在 Cursor 的 Agent 聊天中输入 / 并选择 building-a-beatdeck
Checked against official docs 2026年11月12日 · Source: cursor.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
适用于其他能读取 Skill 的应用。这句话列出了 Claude Code、Codex 和 Cursor 的路径,AI 会按你的应用选择 · 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 2026年11月12日 · Source: code.claude.com, learn.chatgpt.com, cursor.com
其他支持 Agent Skills 的应用(如 Gemini CLI、GitHub Copilot)使用不同文件夹,请查看该应用的指南,再把含 SKILL.md 的文件夹复制过去。
已于 2026年11月11日 对照 Claude Code、Codex 和 Cursor 官方指南核实 skill 文件夹
以下文字来自开发者的 repo,不是 PopNx MCP 的指令
显示前 8,000 个字符 · 在 GitHub 打开完整文件
---
name: building-a-beatdeck
description: Builds a talk as a beatdeck - a fixed 1920x1080 React stage where one click is one beat, with authored motion, presenter view, overview, and an offline self-contained build. Covers turning source material (PPTX, Keynote/PDF export, Markdown, notes) into a content audit and a scene/beat map, choosing or applying a theme, writing scene layers, automatic beats, and verifying every beat with npm run verify. Use whenever the task is creating, porting, restyling, rehearsing or fixing a presentation, talk, keynote, slide deck or conference session that should feel like a live system rather than slides, or whenever the repo contains src/beatdeck/.
license: MIT
---
# Building a beatdeck
A beatdeck is a talk written as code: **8-ish scenes, each made of beats; one click = one beat**. The stage is a
fixed 1920×1080 canvas scaled into any screen with black letterbox. Everything is bundled, so it runs with Wi-Fi off.
Your job is to turn what the speaker wants to say into beats that feel authored — not to reproduce slides.
**Paths.** `references/…` and `scripts/…` below are relative to the folder of this SKILL.md, wherever the skill
is installed (plugin, `~/.claude/skills/`, or `skills/building-a-beatdeck/` inside a deck project). Everything
else (`deck/`, `src/`, `npm run …`) is relative to the deck project.
## 0 · Get a deck repo
If the working directory has no `src/beatdeck/`, scaffold one, clean it and install:
```bash
npx degit borjaperfra/beatdeck my-talk && cd my-talk
npm run init -- --title "<title>" --author "<speaker>" --lang <en|es|…> [--theme light] [--demo]
npm install
```
`init` strips what belongs to the beatdeck repo, not to a talk (the Kernel Panic showcase and its non-MIT
assets, plugin files, beatdeck's README, changelog and CI), renames the package and writes title/author/lang/
theme into the deck. `deck/` becomes a **blank two-scene starter** (standby + title, questions + QR) to build
on; `--demo` keeps beatdeck's 6-scene demo instead, as a reference to read and replace. It also creates
`reference/` (put the source there), `docs/CONTENT-AUDIT.md` and `docs/RUNBOOK.md` from the templates, the talk's
`AGENTS.md` and a CI workflow that builds and verifies every push.
Use `--theme light` when the room is bright or the talk is streamed. Run it once; it deletes itself.
It also resets the demo's `qrUrl` to `TODO` and clears its tagline: set the real QR target only from the source.
Then read `AGENTS.md`. The demo deck (on GitHub in `deck/`, or locally with `init --demo`) is the reference
implementation of every pattern below. `init` removes the heavier example from your project; read it on GitHub when you need it:
https://github.com/borjaperfra/beatdeck/tree/main/examples/kernel-panic — a 47-beat production talk. Look at
`index.tsx` (custom chrome, `isCut` + `Overlay` for a cut to black), `deck/timeline.ts` (every kind of automatic
beat), `architecture/` (one SVG world spanning several scenes) and `simulation/` (a Canvas 2D particle layer).
The talk lives in `deck/`. Never edit `src/beatdeck/` for one talk; if the engine truly lacks something, add it
generically there and say so (`npm run upgrade` replaces engine files; edits are backed up, not merged).
**Existing talk project?** If `.beatdeck.json` exists, `npm run upgrade -- --dry-run` shows whether a newer engine
is available; upgrade only when the user agrees, then `npm install && npm run verify`, and merge any
`*.beatdeck-new` file it reports.
## 1 · Collect the source
Ask for (or find) the source material and the facts you must not guess: title, speaker, event, date, the QR
target URL, real screenshots/logos/photos. Put originals in `reference/` (read-only from then on).
- **PPTX**: `python <skill folder>/scripts/extract_pptx.py deck.pptx reference/source` → one Markdown file with
every slide's text, speaker notes and media, plus the media files. (Inside a scaffolded project the skill
folder is `skills/building-a-beatdeck/`.)
- **PDF / Keynote**: export to PDF and read it page by page; ask for the original images.
- **A written script (Markdown, a doc, notes)**: use as is, and read it by kind of content:
- **narration** (prose, `>` quotes: what the speaker *says*) → the presenter view (`source` / `note`). On
stage it becomes a short statement in the speaker's own words — condensed, never reworded into new claims;
- **facts** (bullets, figures, names) → on stage, exactly as written (same numbers, same rounding, same "~");
- **code blocks, commands, formulas, output** → on stage character for character (`Terminal`, or a
`data-exact` element — see step 6), never retyped from memory;
- **stage directions** ("on screen…", "show how it grows…") → instructions for you, not text to display.
**Nobody to ask?** If you are working without the speaker, decide, record each decision in the audit's
"Decisions taken on the speaker's behalf", and list them in your hand-off. Keep the neutral theme and its
accent unless a brand is given — choosing a palette is an art direction, not a default.
Real artefacts beat redraws: if the source shows a tweet, a logo, a product photo or a meme, use the real file.
Never invent quotes, numbers, tweet text, URLs or logos. Missing facts become explicit `TODO`s and questions.
## 2 · Content audit (before any code)
Write `docs/CONTENT-AUDIT.md` from `references/content-audit-template.md`: every source slide → the beat(s)
that carry it, what is kept, restored or deliberately cut, and open questions. This is what lets you say
"nothing was lost" later. Show it to the user when it changes what the talk says.
## 3 · Beat map
Turn the audit into `deck/scenes.ts` (`SceneDef[]`). Rules (details in `references/beat-model.md`):
- A **scene** is a mode of the talk (a stage world), not a slide title. 5–9 scenes is typical; keys 1–9 jump
to scenes, so more than 9 leaves the rest reachable only by stepping or the overview (O).
- A **beat** is one idea landing. If two things must appear on separate clicks, they are two beats.
- `auto: true` marks a beat whose motion runs by itself and takes noticeable time (a boot sequence, a counter,
a cascade of warnings, a typed command). The presenter flags it and `verify` waits for it. Everything else
waits for the click. Two ways to implement it — pick the lighter:
- **self-contained** (a `Typewriter`, a `CountUp` that restarts on entry): no `timeline` needed;
- **state-driven** (other layers react to its progress, a later beat depends on it, or it advances by itself):
a `Live` field + `initialLive` + `timeline` in `deck/timeline.ts`. See `references/beat-model.md`.
- Fill `ref` (where it comes from), `source` (what the source says) and `note` (speaker cue) — the presenter
view shows them; the audience never does.
- Keep the first scene's first beat as a calm standby screen: the talk starts on the first click. On a dark
theme it can be pure black (`isBlack`); on a light theme use the theme background — anything drawn on a
forced-black standby must not use theme ink colours.
## 4 · Visual direction
Default to the neutral theme (`themes/neutral.css`): near-black stage, off-white type, one accent, hairlines,
huge editorial type, negative space. Bright room, daylight or a stream → `themes/light.css` on top of it
(`init --theme light` does this). If the speaker has a brand, change tokens (`--accent`, fonts, `--bg`) in a
theme file — do not invent a new art direction, decorative labels, stickers, gradients, rounded cards or a
dashboard look. If a design reference exists (Figma, a v3 HTML prototype, a brand book), it decides *how it
looks and moves*; the source decides *what is said*. Record conflicts in the audit.
Fonts must be local: `@fontsource-variable/*` packages or `.woff2` files in the deck, plus their specs in
`fonts` of the deck definition so the first frame never shows a fallback face.
## 5 · Build the scenes
Read `references/authoring.md` and提问、分享技巧或报告无法使用。按“有帮助”数量排序。