rossark-motion-skill: animate drawn illustrations with a Claude Code skill
rossark-motion-skill is a Claude Code skill for animating drawn illustrations in React Native and SwiftUI, with a small motion engine, live-face templates, and tools that measure reference video frame by frame.
What is rossark-motion-skill?
rossark-motion-skill, in the repository rossark0/rossark-motion-skill, is a Claude Code skill that gives an AI coding agent a method, code templates, and measurement tools for animating drawn illustrations in React Native and SwiftUI.
The problem it addresses is animation that looks like a slideshow. The README says an agent asked to bring a character to life will often swap one drawing for another, stop dead at every pose, move everything in sync, and make blinks so short that they flicker. The skill lists those mistakes and how to fix them, so a drawing can breathe, look around, blink, shift its weight, and act without freezing or jumping.
At its centre is a single clock for a whole scene. Moving parts turn around their real pivots, trailing parts follow a beat later, idle life never stops, poses are joined by smooth paths, amplitudes stay small, and curves stay strong. When a reference video is available, the skill measures its motion frame by frame before building, because the detail that makes a reference feel alive is often too small to see with the eye alone.
The skill covers characters and creatures, plants, cloth, fire, smoke, water, weather, lights, machines, icons, logos, charts, and whole scenes. The README says it started as the method the maintainer used to bring the illustrations in an app to life, and grew into a way to animate anything drawn.
Who is rossark-motion-skill for?
rossark-motion-skill is for designers and engineers who want to bring their apps and drawings to life with an AI coding agent, working in React Native or SwiftUI.
It fits readers who already have a drawing and want an agent to animate it. The live-face templates expect the reader to supply the drawing geometry and the colours, and the skill workflow starts by reading the project motion notes and the art file header comments before changing anything.
For React Native, the engine template is written for react-native-reanimated 3/4, and the live-face template uses react-native-svg with react-native-reanimated 3/4. For SwiftUI, the engine and live-face templates state iOS 17+ and macOS 14+. The skill platform table also maps the same engine to the web, using the same math on requestAnimationFrame or a Motion value, while the repository template folders are for React Native and SwiftUI.
The measurement tools need ffmpeg plus Python standard library, and the skill says to run Python with -I. The README gives the demo-rendering commands for macOS 14+.
Quick facts
- Platforms with templates: React Native and SwiftUI; the skill platform table also maps the engine to the web.
- Minimum versions stated in the templates: iOS 17+ and macOS 14+ for SwiftUI; react-native-reanimated 3/4 for the React Native engine.
- License: MIT License, copyright 2026 rossark0.
- Cost: the license grants permission free of charge, subject to including the copyright and permission notices in copies or substantial portions.
- Repository language reported by GitHub: Swift.
- GitHub stars: about 860 as of November 18, 2026.
- Releases: no releases were listed by the GitHub API as of November 18, 2026.
What stands out
SKILL.md covers principles, recipes for every kind of subject, live faces, full-body acting, scenes that tell an action, motion in an interface, pitfalls, working from a reference video, and a workflow.
- A small motion engine with loop and one-shot clocks, track and flow keyframe reading, and helpers named wave, loop, bell, seg, and seeded, plus named curves for travel, opening, and leaving.
- Live faces that glide, blink, and smile: the live-face templates move pupils, eyelids, cheeks, and mouths through keyframe tracks, using geometry and colours supplied by the reader.
- Full-body acting built from flowing poses, idle layers, follow-through delayed from a parent part, weight shifts, landings that squash and keep volume, and props that enter already moving.
- Whole scenes that tell one action: a one-shot clock, beats written as time windows with seg, overlapping beats, and a hold on the final frame with only idle loops running.
- Reference-video tools that measure before building: contact sheets at full frame rate, frame-by-frame edge tracking for motion as small as a 2% jiggle, and exact colour sampling from video frames.
How do I install rossark-motion-skill?
You install rossark-motion-skill by running one command from the project README.
From the project README, the command is:
npx skills@latest add rossark0/rossark-motion-skillThe README then points to SKILL.md as the skill itself, MOTION.md as a notes file to copy next to the project art, the React Native and SwiftUI engine and live-face templates, the three measurement tools, and the demo folder. The skill says to copy the engine template into the project, unless the project's own notes name a different engine, and to copy MOTION.md next to the art when a project has no motion notes, filling in its second part as the project is learned.
The README also describes rendering the demos again on macOS 14+ with the SwiftUI templates and the demo files, through an offline renderer in the demo folder that turns scenes into MP4 and GIF output.
Try it with your AI. Try it with your own drawing and agent, using only what the README describes.
Help me animate one drawing I already have, using only rossark-motion-skill and the
official commands and interfaces named in its README. My drawing is [describe the
drawing and its file path]. My project uses [React Native or SwiftUI], and the art lives
at [path]. First, install the skill with the README command npx skills@latest add
rossark0/rossark-motion-skill, then read SKILL.md and copy MOTION.md next to my art,
filling in Part 2 for my project. Copy the matching engine template (motion.ts or
Motion.swift) and live-face template (liveFace.tsx or LiveFace.swift) into my project,
and animate [one main action] plus [two or three idle loops], using one clock, track or
flow keyframes, and the wave, loop, bell, seg, and seeded helpers. Show me every risky
step first and wait for my confirmation before changing any file. Do not install
anything from unknown sources, do not touch credential files, and do not delete
anything. Tell me what moves, what was left still on purpose, and what has not been seen
on a real device.What are rossark-motion-skill's limits?
rossark-motion-skill's limits are the platform versions its templates state, the checks its drawing rules require, and a rule against publishing a copy of someone else's art.
The SwiftUI templates state iOS 17+ and macOS 14+, and the README gives demo rendering for macOS 14+. The React Native engine header names react-native-reanimated 3/4.
Fast moves need enough frames: the skill says anything under about 100 ms flickers at 30 fps, and to render and export at 60 fps when the platform allows. Reduced motion means still: loops freeze on frame 0 and one-shot scenes show their final frame, so both frames must read on their own. Clocks must pause when off screen, hidden, or in the background, and in SwiftUI a TimelineView redraws every frame while it runs, so the skill says to pause it and keep one per scene rather than one per part.
Drawing errors show up in motion. The skill warns that two anti-aliased edges on the same line can leave a hairline seam, overlapping parts in one path can punch holes where directions differ, and a lid clipped to the exact eye white can leave a glasses ring around closed eyes. It also says many items should be drawn in one SVG or Canvas from the same clock, never one animated view each, and that unique SVG ids matter because clip paths are global to the screen.
For reference video, the skill says measuring and copying a drawing exactly is fine for a private test, but a published demo, repository, or product needs original characters and art, because brand mascots and illustrations are trademarks and copyrighted, and MOTION.md says not to publish a copy of someone else's art. The repository contains no separate security or privacy document in its file tree, and the MIT license provides the software as is, without warranty of any kind.
Project-specific rig limits are not fixed by the skill: MOTION.md asks the reader to record what their own art cannot do, and project rules recorded there override the skill defaults.
Related on claudefox
- Dashboard Design Skill: a Claude Code skill for clean, honest dashboards
- antialgorithm: a Claude Code skill that moves creative work past the first average pick
- motlab: video as code with canvas scenes, a review viewer and agent skills
- stack-diagram: an unofficial agent skill that turns one sentence into an architecture diagram with official logo files
Problems and feedback
Report problems and give feedback on the repository Issues page at https://github.com/rossark0/rossark-motion-skill/issues.
Source: rossark-motion-skill repository, README, SKILL.md, MOTION.md, LICENSE, motion.ts template, Motion.swift template, liveFace.tsx template, LiveFace.swift template, Releases, Issues
This story first appeared in the claudefox daily briefing for November 19, 2026.