LottieFiles motion-design-skill: 12 Motion Design Principles for AI Agents
An open-source skill pack that injects UI timing, easing curves, and Disney's 12 principles of animation into over 40 AI coding agents including Claude Code and
motion-design-skill, open-sourced by LottieFiles, is a structured prompt and guideline skill pack designed to teach AI coding agents how to think like motion designers. It installs with a single command across more than 40 coding agents, including Claude Code, Cursor, Codex, and GitHub Copilot.

Image source: LottieFiles 쨌 CC BY-SA 4.0 쨌 https://creativecommons.org/licenses/by-sa/4.0 쨌 https://commons.wikimedia.org/wiki/File:ThorVG_Icon_Primary.png
When prompted to create UI animations, conventional AI coding assistants often produce generic, lifeless code such as transition: all 0.3s ease or trigger chaotic, unsynchronized movements across adjacent elements. This repository addresses that issue by structuring the agent's decision-making process, ensuring it computes timing, easing curves, choreography, and emotional intent before writing any animation code.
What Agent Skills Are and Why Motion Guidance Matters
In AI-assisted software development, an 'agent skill' functions as a modular instruction set that supplies specialized domain knowledge and concrete execution rules to an autonomous assistant. Rather than forcing developers to repeatedly paste long system prompts into every chat session, skills activate contextually whenever relevant tasks are detected.
motion-design-skill organizes its motion director framework across four distinct layers:
- Core (
SKILL.md): A concise master instruction file under 500 lines containing an eight-step checklist, motion personality archetypes, timing and easing lookup tables, property selection rules, and quality assurance rubrics. - Philosophy (
director/): In-depth guides that adapt Disney's classic 12 principles of animation—first documented in 1981—to contemporary user interfaces. This section details how anticipation, slow in and slow out, staging, and secondary action translate directly into digital interface states. - Patterns (
patterns/): Practical, production-ready recipes covering entrance and exit choreographies, interactive state feedback (success, error, loading, hover), ambient background motions, and multi-element sequence staggering. - Reference (
reference/): Millisecond-level duration tables, mathematical easing curves, property communication guides, and an animation troubleshooting directory for identifying performance bottlenecks.
Once the skill is mounted, asking an agent to "animate a set of dashboard cards entering the viewport" prompts it to automatically compute staggered entrance delays, choose suitable easing curves, and coordinate visual hierarchy, rather than revealing all elements at once.
Implementation-Agnostic Design and Installation
A key technical characteristic of the repository is its implementation-agnostic philosophy. It does not dictate a particular rendering engine or UI framework. The directional principles apply universally, whether your project runs on native CSS transitions and keyframes, React-based Framer Motion, the GSAP interaction suite, Lottie vector runtimes, or physics-based spring models.
The skill installs directly via the skills package manager CLI:
npx skills add LottieFiles/motion-design-skill
Executing this command places the structured guidelines into your agent's configuration directory. The skill then activates seamlessly when handling natural language requests such as:
Add a playful entrance animation to these card components
Create a loading -> success -> error state transition for this form submit button
Choreograph the entrance sequence for this dashboard: hero chart first, then stats cards, then sidebar
Review my animation code and suggest improvements based on motion design principles
Instead of simply dumping code, the agent explains its motion reasoning alongside the implementation—clarifying, for instance, why a 100ms scale-down feedback was chosen for a button to ensure instantaneous tactile feedback.
Practical Considerations and Limitations
motion-design-skill is not an executable runtime library or animation engine. It consists entirely of Markdown-based guidelines and heuristic rules designed to refine an agent's reasoning. Consequently, your project must already include the target animation framework or styling solution (such as CSS, Tailwind CSS, Framer Motion, or GSAP) to execute the generated code.
Furthermore, in complex visualization scenarios such as 3D canvas rendering or heavy particle systems, timing recommendations generated by an LLM still require human developer validation. Developers should ensure that the generated animations favor GPU-accelerated properties (transform and opacity) rather than trigger expensive browser layout recalculations, and verify that accessibility safeguards such as prefers-reduced-motion media queries remain strictly respected.
Maintained officially by LottieFiles under the MIT License, the repository has surpassed 3,000 GitHub stars, serving as a prominent reference model for embedding specialized design capabilities into AI developer tooling.
Sources
- GitHub Repository: LottieFiles/motion-design-skill