Japanese Developer Releases 120 AI Animation Terminology Library
Japanese developer Shohei Ohtaki has launched the Motion Prompt project, organizing professional terms from animation, photography, and interface design into a name library directly callable by AI, containing a total of 120 entries, each accompanied by an interactive dynamic demonstration.
The project is divided into 5 areas, each with 24 items: motion graphics, camera movement and transitions, UI component names, landing page and web effects, and photo lighting and styles. All demonstrations are based on pure time function implementations in Canvas 2D, requiring no external dependencies and can loop seamlessly.
Users can install the project as a skill for Claude Code or Codex by running npx skills add showheyohtaki/motion-prompt. After installation, they can open the gallery by issuing the command "open the motion-prompt library," select effects, and click send, allowing the AI to receive corresponding prompts and reference implementations.
Each entry includes a three-digit number, English and Japanese names, a one-sentence definition, and example sentences for requesting from AI. Developers can directly reference numbers like "use #012 and #045" or provide the AI with demonstration links and specifications by copying selected results.
The project is open-sourced under the MIT license and also provides a make-your-own directory containing prompts and rendering scripts, allowing other developers to create similar name libraries in their own fields and languages. The repository has received over 180 stars on GitHub.
In terms of market mechanisms, developers willing to structure professional terminology gain priority access to AI tools, while ordinary users reduce description costs by referencing numbers. AI agents and coding tool providers benefit from reusable visual specifications, facing efficiency pressures compared to traditional workflows relying on long text prompts.
The project also includes a static API and command-line tools, supporting calls in environments without skills by pasting prompts or directly referencing numbers. All demonstrations are stored in pure function form, allowing the same code to be previewed in real-time or rendered frame-by-frame as video.
Source: Public Information
ABAB AI Insight
Shohei Ohtaki has previously released projects like motion-explainer, continuously transforming specialized terminology in specific fields into AI-understandable animation interpreters. The Motion Prompt continues this path by digitizing the existing vocabulary system in animation and design, lowering the access threshold through GitHub Pages and skill installation mechanisms, representing a series of practices that convert tacit knowledge into callable interfaces.
In terms of capital pathways, the developer has open-sourced the personally organized terminology library under the MIT license and distributed it through skill markets and prompt installation pathways, allowing AI agents to prioritize this library when handling visual tasks. Resources shift from decentralized natural language descriptions to numbered references, motivated by the goal of improving AI generation consistency and reducing the time cost of repeated debugging.
This is analogous to the early open-source icon libraries and design systems that standardized visual elements. Currently, we are at a stage where AI coding agents require precise visual specifications. Similar to how Tailwind or Radix reduce development friction through named components, Motion Prompt extends the same logic to dynamic effects and camera language.
Structurally, this represents a technological substitution, where professional terminology shifts from relying on human experiential tacit knowledge to being directly indexable resources by AI. The mechanism is that AI faces high uncertainty with natural language descriptions, while calls to clear numbers and reference implementations are more stable, thus granting the party with a structured knowledge base higher control over the generated results.
ABAB News · Cognitive Laws
Giving AI a name is better than a lengthy description
Only after structuring professional terminology can it be reliably called by AI
Organizing existing knowledge is more valuable than reinventing the wheel