Five installable SKILL.md workflows and a curated catalog of GSAP, Motion, Remotion, Rive, Lottie, and 3D tools for AI coding agents in Claude Code, Codex, Cursor, and Copilot.

Every entry links to its primary source and carries a dated pulse. Companion FrankX lists sit at the end, never in front.

Doctrine: static hierarchy first. Animate only to clarify state or narrative. Design the reduced-motion path — don't just disable it. Verify on real devices.

Five skills ship in skills/ as open-standard SKILL.md folders, verified on fresh checkouts for Claude Code and Codex.

Verify on a fresh checkout:

npx --yes skills add frankxai/awesome-motion-design-agent-skills --list
npx --yes skills add frankxai/awesome-motion-design-agent-skills -a claude-code -y
npx --yes skills add frankxai/awesome-motion-design-agent-skills -a codex -y
npx --yes skills list

Scan any third-party pack with NVIDIA/SkillSpector before it reaches a live profile.

Five first-party skills, one per job. Read the SKILL.md, then install only what the work needs.

New here? GETTING_STARTED walks discovery → read → install → verify.

Operators get leverage from about 5–7 named workflows, not bulk dumps. Method: earned-skills index · safety gate.

Do not install unsigned ZIP/S3 skill blobs or mass-dump skill collections into a live profile.

The 1human client serves reference discovery rather than rendering: public previews need no account, while recipes, publication and private preferences require owner-approved credentials. The client/examples are MIT; the hosted service, brand and private data are not. Reference licenses remain per creation. Basic participation is free, with a capped beta; this is not a universal importer or evidence of adoption.

Valmera’s website-demo skill covers public visitor flows or approved screen recordings. Editing runs on the hosted service and requires authentication and a paid plan; the MIT license covers the plugin and skill source, not the hosted editor. The skills require an approved scope before billable work and a reviewed preview before final export. Worked trim-and-reframe export · current release.

mindmap
  root((Curated agent capability))
    Strategy
      fit and scope
    Governance
      provenance and license
    Talent
      human review
    Technology
      tools and integration
    Data
      evidence and memory
    Ethics
      safety and disclosure

Companion catalogs. The third-party projects above are this list's primary value.

Agents & systems — agent operating systems · hermes agents · hermes agent skills · automation · AI CoE

Create — design · motion design · music · gamification

Capital — agentic income · investor · wealth · payments

Mind & cosmos — mind · manifestation · cosmos

See CONTRIBUTING.md for the full pull-request checklist. Open a PR with a primary URL, one-sentence distinct value, current maintenance evidence, license posture, and relevant safety or reduced-motion caveat. New skills must pass skills-validate (frontmatter name matches folder, description ≤ 1024 chars, no private paths) and fresh-checkout installation in Claude Code and Codex.

Repository metadata (stars, SPDX license, archived state, last push) was pulled from the GitHub API on 2026-10-24. "Custom license" means GitHub returned no standard SPDX identifier — read the license before adoption. Counts are dated discovery signals, not rankings. Links are checked on every PR and monthly by Lychee. Nothing here is financial, legal, medical, or safety advice.