Motion Graphics Skills for AI Coding Agents
56 open-source skills that teach your AI coding agent to make motion graphics, animation, and video — kinetic typography, data-driven charts, explainers, TikTok / Reels, 3D & WebGL, math animation, generative illustration, code-drawn canvas animation, transparent lower thirds, b-roll for talking-head videos, AI presenter films, and web motion. Nineteen installable packs. Install one, and your agent just knows how.

Motion is the hardest thing to get right from a prompt: timing and easing, safe areas, frame-accurate rendering, on-brand type and color, a loop that actually loops. These skills encode that craft as agent-native workflows — each one teaches the agent how a professional would build it, with a deliver-and-verify loop (render a frame → screenshot → check) so it inspects its own output instead of hoping.
They're the open-source craft behind iart.ai, the AI motion agent. Bring them into the agent you already use — or let iart.ai run the whole pipeline for you.
The 19 packs
Each pack is its own repo, aimed at one audience. Install only what you need.
Browse every pack and skill, with install commands, at iart.ai/skills.
Install
Each pack installs on its own:
npx skills add iart-ai/tiktok-video-skills
npx skills add iart-ai/data-animation-skills
# …or any pack above
Or add one as a Claude Code plugin:
/plugin marketplace add iart-ai/web-animation-skills
Once installed, skills are auto-discovered by Claude Code, Cursor, Codex, and 40+ agents — and activate on their own when your prompt matches the work.
Or let iArt run it

Made in iArt with DeepSeek, each from a one-line prompt.
These skills are the craft inside iArt, the strongest DeepSeek animation agent and the cheapest Claude animation agent. On Claude, prebuilt scaffolding, context design and tool design reach the same quality at about 1/5 of the cost of running the model yourself (average of 10,000+ ablation runs on Claude). Made for the latest models: DeepSeek 4.1, Claude Fable 5.2, Opus 5.5, Sonnet 5.5, Haiku 5.5 and Gemini 4 Argon. Voiceover, image generation and AI video come out of the same credit. See pricing →
Or call iArt from your agent (MCP)
iArt also runs as a remote MCP server (beta), so the agent you already code with can hand a whole video to iArt: describe it in chat, get back an MP4 and a share link a few minutes later. Pass your logo or screenshots as files, revise it in plain words. Part of the Pro, Ultra and Max plans; create an API key in the app (account menu → API).
# Claude Code
claude mcp add --transport http iart https://app.iart.ai/api/mcp --header "Authorization: Bearer YOUR_API_KEY"
# Codex
export IART_API_KEY=YOUR_API_KEY
codex mcp add iart --url https://app.iart.ai/api/mcp --bearer-token-env-var IART_API_KEY
In the Claude app (claude.ai or desktop) and ChatGPT there is no key: add https://app.iart.ai/api/mcp as a custom connector (Claude) or a custom MCP server with OAuth (ChatGPT), sign in to iArt, then ask for a video in the chat. Step-by-step with screenshots: Claude · ChatGPT.
Cursor, VS Code and the REST API: iart.ai/docs/mcp.
What's a skill?
A skill is a self-contained folder (SKILL.md + references/) that teaches an agent one motion-graphics workflow — kinetic typography, beat-synced cuts, frame-accurate data charts, accessible web animation, and so on. No model fine-tuning, no plugin runtime: just the know-how, in a format every major agent already reads.
Skills that produce a visual artifact ship a deliver-and-verify loop plus a small scripts/ toolkit (freeze a frame, tile a contact sheet, probe the encoded MP4), so the agent confirms its own output — web skills render a standalone HTML, video skills render via Remotion or Manim.
Built for every kind of motion
Short-form vertical video • explainer and educational clips • data-driven infographics and charts • product and e-commerce demos • brand and ad creative • kinetic typography and titles • editorial map animation • logo and brand-system motion • GSAP / SVG / Lottie web animation • Three.js / WebGL / shaders • Manim math animation • Remotion programmatic video • zero-asset canvas animation drawn frame by frame in JavaScript • transparent lower thirds and stream overlays (WebM alpha, ProRes 4444) • b-roll for talking-head videos, timed to the words.
License
MIT — use them freely.
Built by iart.ai, the AI motion agent — describe it in a prompt, or point it at a CSV or brand kit, and get editable, on-brand motion graphics with exact text and numbers.