motion-prompt README

项目的 README,来自 github.com/showheyohtaki/motion-prompt。合并为一个 markdown 文件:README.md。

Motion Prompt

English | 日本語

日本語の説明は README.ja.md にあります。 中文用户:所有名称一览都有简体中文和繁體中文版本,见 packs/。 En español: todas las listas de nombres están en español en packs/. The name lists are in English, Japanese, Spanish, Simplified Chinese and Traditional Chinese (see packs/); the gallery is in English and Japanese. For another language, give make-your-own/prompts/en/01-translate.md to a coding AI.

A library of names for the things you ask an AI to make. Each name comes with a moving sample, a one-line meaning and an example of how to ask for it. Nine sets are in it now, 24 names each and 216 in all: motion-graphics moves (motion), camera moves (camera), UI component names (ui), landing page and web effects (lp), photo looks, lighting and styles (photo), motion-graphics moves, part 2 (motion2), landing page and web effects, part 2 (lp2), camera moves and transitions, part 2 (camera2), and art styles for image generation (artstyle). It also comes with the prompts to make the same kind of library for your own field, in your own language.

Motion graphics Camera moves
Motion graphics names 01–08 Camera move names 01–08

Why I made this

I posted a few short videos that put names to motion-graphics moves and camera moves, and over the next few days a lot of comments came in. One of them said, in essence, that what had been holding them back was not a lack of imagination. It was not knowing what this kind of movement is called.

Reading that, I thought: if people who know their field, in many languages, laid out the "common language" you need to work well with AI as a library you can look through, it would be really useful. So I tried making that as a GitHub repository.

This is not one style to copy. It is a place where you can use the prompts and instructions here to make motion graphics like these, and to make the same kind of name samples for your own field, in your own language, with whatever look you like.

Another comment said: "this is how you become a creative director". Exactly. I think it would be a wonderful thing if there were more creative directors all over the world, doing great work together with AI.

Shohei Ohtaki

The names in here now

Numbers Pack Names Lists
001–024 motion: motion graphics moves 24 English · 日本語 · Español · 简体中文 · 繁體中文
025–048 camera: camera moves and transitions 24 English · 日本語 · Español · 简体中文 · 繁體中文
049–072 ui: UI component names 24 English · 日本語 · Español · 简体中文 · 繁體中文
073–096 lp: landing page and web effects 24 English · 日本語 · Español · 简体中文 · 繁體中文
097–120 photo: photo looks, lighting and styles 24 English · 日本語 · Español · 简体中文 · 繁體中文
121–144 motion2: motion graphics moves, part 2 24 English · 日本語 · Español · 简体中文 · 繁體中文
145–168 lp2: landing page and web effects, part 2 24 English · 日本語 · Español · 简体中文 · 繁體中文
169–192 camera2: camera moves and transitions, part 2 24 English · 日本語 · Español · 简体中文 · 繁體中文
193–216 artstyle: art styles for image generation 24 English · 日本語 · Español · 简体中文 · 繁體中文

Each list gives the number, the name, what it means and an example of how to ask an AI for it, in English, Japanese, Spanish, Simplified Chinese and Traditional Chinese. The same names move in the gallery.

Each pack has its own look. The motion and motion2 packs are dark navy with four colors. The camera and camera2 packs are light gray with one blue. The ui pack is warm off-white with black and one blue, and a cursor sets each part moving. The lp and lp2 packs draw a white browser window on a pale gray desk, with a vermilion accent and blue for the pictures on the page. The photo pack has no shared palette: each card is painted in the colors of its own scene, and the effect fades in and back out so you can compare. The artstyle pack is the one pack with pictures instead of drawings: the same base sketch was given to an image model and redrawn in each style, and a diagonal wipe turns the sketch into the style. That is on purpose. What a pack has to get right is the names, not a house style.

Three ways to use it

1. Open the library, pick, and send your picks to your AI

With the motion-prompt skill installed in Claude Code or Codex, asking is enough to open this gallery. You do not need to remember names or numbers. When you are not sure what something is called, have the library opened, the way you would look a word up in a dictionary.

npx skills add showheyohtaki/motion-prompt
  1. Ask your AI:
    Open the motion-prompt library.
    
  2. The gallery opens in your browser. Press "Pick" on the samples you want, as many as you like. The note field at the bottom of the screen takes a short line such as "for the opening title".
  3. Press "Send to AI" and your picks reach the AI. Go back to it and say what you want done:
    Edit the opening of this video like this.
    

The agent uses only what you picked. It fetches their specs and reference code and builds them into your video or site the way your project works. Name a kind, as in "show me the camera moves", and the gallery opens filtered to that field's packs (part 1 and part 2 together where there is one).

For an AI without the skill, pick in the gallery, press "Copy picks" at the bottom of the screen and paste the text into the chat. It includes the demo and spec URLs.

When you know the numbers, you can skip the gallery and ask by number:

Use the motion-prompt skill with #012 and #045 to make this product video look professional.
Replace the part where the title appears with #005 from motion-prompt. Keep the length and the transitions before and after it.
Turn the camera move in motion-prompt #032 into an English prompt for Veo. The subject is a coffee cup on a desk.

Other ways to install, and what the agent does, are in Skill reference below.

2. Look up the name and say it to your AI

Open a list or the gallery, find the thing you have in mind, and use its name in your prompt. Adding how long, how strong and what moves makes the result closer to what you want.

Bring the title in with a mask reveal from below, 0.6 s, each line slightly after the one before.
Slow dolly in toward her face; move the camera, do not zoom.

No install is needed for this. The names work in a coding AI, a video-generation model or a brief to a person.

3. Make a version for your own field or language

make-your-own/ has copy-and-paste prompts for a coding AI, a script that renders the videos, and the two finished examples (HTML source, three videos and a cheat-sheet image each).

Goal Prompt
The same videos in another language 01-translate.md
The names of your own field, with your own look 02-your-own-field.md
The one-page cheat-sheet image 03-cheatsheet-image.md

The field does not have to be video. Lighting, sound, UI, photography, cooking, architecture: any field where knowing the name is what lets you ask for the thing.

Add the names of your field

If you know a field well, please add its common language here, in your language. CONTRIBUTING.md explains how to add a pack or add a language to an existing pack, and what a name needs before it goes in. A coding AI can do most of the building with the prompts above.

Original posts

The videos this started from (X, in Japanese):

The set from the first post (names of animations for websites and landing pages) is not in this repository yet. The lp pack is a different set. Some names overlap with that post (parallax, scroll snap and others), but the selection and the samples were made new for this repository. The ui pack is based on "UIパーツ名24" (24 UI part names), posted on X on October 10, and photo was also made new for this repository.

Skill reference

Node.js 18 or later is required.

Install

With a prompt. Paste this into Claude Code or Codex:

Read https://github.com/showheyohtaki/motion-prompt and install the motion-prompt skill in this environment. When it is installed, tell me how to use it in three lines.

With a command. This uses the skills installer, which works with Claude Code, Codex and other agents that read Agent Skills:

npx skills add showheyohtaki/motion-prompt

As a Claude Code plugin.

/plugin marketplace add showheyohtaki/motion-prompt
/plugin install motion-prompt@motion-prompt

By hand. Copy the skills/motion-prompt folder to one of these places:

Location Scope
~/.claude/skills/ Claude Code, all projects
~/.agents/skills/ Codex and others, all projects
.claude/skills/ in a project Claude Code, that project only
.agents/skills/ in a project Codex and others, that project only

Install steps for agents

This section is for an AI agent that received the prompt in "With a prompt".

  1. Check that Node.js 18 or later is available with node --version.
  2. In the user's project folder, run the command below. If it asks which agent to install for, choose the one you are.
    npx skills add showheyohtaki/motion-prompt -y
    
  3. If step 2 is not possible, fetch the repository and copy the folder. Copy to .claude/skills/ if you are Claude Code and to .agents/skills/ otherwise (or to ~/.claude/skills/ or ~/.agents/skills/ for all projects).
    git clone --depth 1 https://github.com/showheyohtaki/motion-prompt.git /tmp/motion-prompt
    mkdir -p .claude/skills
    cp -R /tmp/motion-prompt/skills/motion-prompt .claude/skills/
    
  4. Confirm the install. Replace <skill-dir> with the path where the skill was installed (the one with SKILL.md in it). It worked if the data source and the element count are printed.
    node <skill-dir>/scripts/motion-prompt.mjs info
    
    Exit code 3 means the data source could not be reached. Get network permission and run it again.
  5. Read <skill-dir>/SKILL.md and tell the user how to use the skill in three lines. For example:
    • Say "open the motion-prompt library" and the gallery opens. Press "Pick" on what you want, then "Send to AI".
    • Then say what to do with them, for example "edit the opening of this video like this".
    • If you know the numbers, you can also ask directly: "use #012 and #045".

Do not change existing settings files or other skills.

What the agent does

How it works

Three files make up the skill: the instructions (SKILL.md), notes on moving an element into each kind of video project (references/porting.md), and a Node script that looks elements up (scripts/motion-prompt.mjs, standard library only).

Command What it does
pick / pick --pack camera opens the gallery and prints what arrives with "Send to AI": numbers, names, specs and your note. Waits up to 15 minutes
search typewriter text / search 文字 登場 finds elements for a scene (English or Japanese words)
list / list --pack motion an overview of all packs, or one pack listed element by element
get 12 #045 prints the spec and then the code for those numbers (8 at most in one call)
kit motion the shared code of a pack
prompt 12 the copy-paste prompt for an element
info shows where data is coming from, how much there is, and where it is cached

The script makes GET requests for the text files below and nothing else. Nothing from your project is uploaded. pick also opens a temporary listener on your own machine (127.0.0.1) that receives only the numbers and note sent by "Send to AI" in the gallery.

  1. https://showheyohtaki.github.io/motion-prompt/api/v1/
  2. if that is unreachable: https://raw.githubusercontent.com/showheyohtaki/motion-prompt/main/api/v1/
File Contents
catalog.json version, count, packs, and every element's names, meaning, summary, spec and tags
{ja,en}/catalog.tsv the index as tab-separated text (id, pack, category, title, tags, summary)
{ja,en}/e/{id}.txt a single element: attribution line, what it shows, the spec, then its code
{ja,en}/p/{id}.txt the copy-paste prompt for one element
kit/{pack}.js the shared code of a pack

If you clone this whole repository, the CLI reads the local api/v1 first and needs no network. Downloaded data is kept on disk, and the saved copy is used when the source cannot be reached.

Variable Default
MOTION_PROMPT_BASE read from here instead of the default sources: a local folder or a URL, several separated by commas
MOTION_PROMPT_CACHE where downloads are kept; ~/.cache/motion-prompt, or the system temp folder when the home folder is read-only
MOTION_PROMPT_TTL how long a saved catalog is trusted, in seconds; 21600 (6 hours)
git clone https://github.com/showheyohtaki/motion-prompt.git
cd motion-prompt
python3 -m http.server 8000

Open http://localhost:8000/ to see the gallery (index.html). In a clone, node skills/motion-prompt/scripts/motion-prompt.mjs pick --local opens your local copy of the gallery for picking. The file formats are described in FORMAT.md.

License

MIT © 2026 Shohei Ohtaki

Commercial use is allowed, for the code, the videos and the images. When you use an element's code, keep its first line (// Motion Prompt #ID · MIT License (c) 2026 Shohei Ohtaki · …).

To ask for something to be removed or corrected, use the removal or correction form. We act on it as soon as the request is confirmed.

Acknowledgements

The idea of pairing a gallery of live examples with an agent skill comes from Lucian UX/UI Prompt. The CLI's command set and the layout of the static data files follow that project too; see THIRD-PARTY-NOTICES.md.

The name lists in other languages began with translations from @daddydanielt (Traditional Chinese), @herberthcr (Spanish) and @xhonye (Simplified Chinese). Thank you.

Author

Shohei Ohtaki (株式会社デイトラ) / X @showheyohtaki

motion-prompt (showheyohtaki/motion-prompt)

正在搜索数据库…

这是什么?motion-prompt 是一个包含 120 种具名效果的库(24 个动态图形动作、24 个镜头运动、24 个 UI 组件名称、24 个落地页效果和 24 种照片风格),每种效果都配有一个动态示例、一句话释义和一条示例提示词,这样你就能明确告诉 AI 想要什么效果,而不必含糊地描述。
这是可浏览的图库、资源库,还是 AI 智能体技能?三者都是:一个可以自己在浏览器中运行的图库,一组可以在 GitHub 上直接阅读的纯文本名称列表和规格说明,以及一个可安装的“Agent Skill”(包含 SKILL.md 和 Node.js 命令行工具),让 Claude Code 或 Codex 打开图库,并为你选中的效果获取对应规格。
怎样在 Claude Code 或 Codex 中安装这个技能?你可以直接把指定的安装提示词粘贴到 Claude Code 或 Codex 中,运行 skills 安装命令,将它添加为 Claude Code 插件,或者手动把 skills/motion-prompt 文件夹复制到 ~/.claude/skills/ 或 ~/.agents/skills/ 中(也可以复制到项目的 .claude/skills/ 目录)。
运行这个需要安装 Node.js 吗?需要,运行该技能的命令行脚本需要 Node.js 18 或更高版本;脚本本身只使用 Node 标准库,没有任何外部依赖。
图库需要联网才能使用吗?默认情况下,命令行工具会通过网络获取目录和元素数据(缓存约 6 小时),如果无法访问数据源,则以退出码 3 失败;克隆整个仓库后,命令行工具可以改为读取本地文件,从而离线工作。
这个库里的五个分类包分别是什么?motion(动态图形动作)、camera(镜头运动与转场)、ui(UI 组件名称)、lp(落地页与网页效果)和 photo(照片风格、光影与样式)——每类 24 个名称,共 120 个。
作者为什么做这个项目?作者在发布为动态图形和镜头运动命名的短视频后意识到,很多创作者并非缺乏想象力,而是不知道某种运动叫什么名字,于是这个仓库把这些命名词汇整理成一个共享库,方便与 AI 协作。
我能为自己的领域或语言做一个版本吗?可以。make-your-own/ 文件夹中有可直接复制粘贴的提示词(供编程 AI 使用)、一个渲染脚本,以及两个已完成的示例;它适用于任何需要统一命名的领域,例如灯光、声音、UI、烹饪、建筑等,并支持任何语言。
AI 智能体如何使用我的选择?在图库中点击“Pick”和“Send to AI”之后,智能体会先判断你的项目如何生成视频(例如 Canvas、Remotion、CSS、After Effects 等),然后只获取你所选效果的规格和参考代码,并在你的项目中精确还原这些动作。
大小
33.4 MB
星标
2.3K
更新
提交Nov 16
时长
1个月
提交 · 90 天