iart-ai/motion-skills: 50 open-source skills that teach your AI coding agent to make motion graphics, animation & video — kinetic typography, data-viz, explainers, TikTok/Reels, WebGL, Manim. 14 installable packs. By iart.ai, the AI motion agent.

50 open-source skills that teach your AI coding agent to make motion graphics, animation & video — kinetic typography, data-viz, explainers, TikTok/Reels, WebGL, Manim. 14 installable packs. By...

Summary

iart-ai/motion-skills is a curated set of open-source "skills" that teach AI coding agents how to produce motion graphics, animation, and video. Each skill is a self-contained folder containing a SKILL.md plus reference material, and the repo groups 54 of these across 17 separately installable packs (TikTok/Reels, data animation, web animation, WebGL, Manim, kinetic typography, lower thirds, and more). Skills that produce a visual artifact ship a deliver-and-verify loop — the agent renders a frame, screenshots it, and inspects the output before continuing.

What it is useful for

The repo addresses a specific problem: getting AI agents to produce motion that looks professional, not just technically rendered. Motion has constraints that are hard to express in a prompt alone — timing, easing, safe areas, frame-accurate rendering, on-brand typography, and loops that actually loop. These skills encode that craft for the agent.

It's useful when you want a coding agent to:

• Generate short-form vertical video for TikTok, Reels, or Shorts (via the tiktok-video-skills pack, which renders with Remotion).
• Build data-driven charts and explainer clips that render to video with exact numbers preserved.
• Produce web animation (GSAP, SVG, Lottie) shipped as standalone HTML.
• Create Three.js / WebGL / shader work, math animations via Manim, or zero-asset canvas films drawn frame by frame in JavaScript.
• Generate transparent lower thirds (WebM alpha, ProRes 4444) for OBS, Premiere, or DaVinci Resolve.

A small scripts/ toolkit per skill lets the agent freeze a frame, tile a contact sheet, and probe the encoded MP4 so it can check its own work instead of guessing.

How engineers can use it

The repo itself is an index; you install the pack that matches your task, not the whole monorepo. The README documents two install paths:

# Install a specific pack as agent skills
npx skills add iart-ai/tiktok-video-skills
npx skills add iart-ai/data-animation-skills

# Or add one as a Claude Code plugin
/plugin marketplace add iart-ai/web-animation-skills

Once installed, the skills are auto-discovered by Claude Code, Cursor, Codex, and other compatible agents, and they activate when your prompt matches the workflow. Video skills render through Remotion or Manim; web skills render to a standalone HTML file.

A documented limitation: this README only shows installation for npx skills add and the Claude Code plugin command — agents outside Claude Code may need their own skill-loading configuration, so check your agent's docs if npx skills is not recognized. All packs are MIT licensed.