The Agent Skills Registry automatically discovers and indexes npm packages that ship Agent Skills. There's no manual submission process — publish skills in your package and the registry picks them up.
The registry periodically searches npm for packages with the tanstack-intent keyword. When it finds one, it downloads the tarball, extracts every skills/**/SKILL.md file, and indexes the contents. Each new version you publish gets indexed automatically.
Tell your AI coding agent to run:
npx @tanstack/intent@latest scaffold
This walks the agent through domain discovery, skill tree generation, and skill creation. You review at each stage. Skills land in a skills/ directory at your package root — each as a SKILL.md file in its own subdirectory.
npx @tanstack/intent@latest validate
Catches structural issues, missing frontmatter, and broken source references before you publish.
Add "tanstack-intent" to the keywords array in your package.json:
{
"keywords": ["tanstack-intent"]
}
This is how the registry finds your package on npm.
npm publish
The registry discovers your package on its next sync cycle. Your skills, version history, and download stats appear on the registry automatically.
Skills derived from docs drift when docs change. Two commands keep them honest:
npx @tanstack/intent@latest stale
Flags skills whose source docs have changed since the skill was last updated.
npx @tanstack/intent@latest setup-github-actions
Copies CI workflow templates into your repo so validation and staleness checks run on every push. Catch drift before it ships.
If you use a library that doesn't ship skills yet, the best path is to open an issue on that library's repo pointing them here. The maintainer is the right person to author and own skills for their tool — they know the intent behind the API better than anyone.
You can also point them to the Agent Skills spec and the TanStack Intent overview for context.