How AI fits into pota — the skills the package ships to help you build with AI assistance, and an honest account of how AI was used on the documentation and the project itself.
pota ships two skills inside the npm package, under
node_modules/pota/.claude/skills/. AI coding tools that follow the
skill convention discover them automatically once pota is a dependency
— there is nothing to install or configure. Both are consumer-facing: they help you build with
pota and contain nothing about developing pota itself.
pota — write idiomatic pota. The day-to-day conventions for
writing pota: JSX rules (on:click vs onClick, class not
className, reader-function children), the signal API
(read / write / update), the built-in components, the store
layer, and use/* plugins. Reach for it when writing or reviewing
components, fixing a pota bug, or creating examples.breaking-changes — migrate across versions. A catalogue of
pota's breaking changes with before/after migrations: the signal
tuple removal, use:* behaviour plugins becoming use:ref
factories, removed and renamed exports, the pota/web →
pota/components rename, and more. Reach for it when bumping the
pota dependency or fixing an app that calls an API pota changed.Each migration is tagged with the version it shipped in, so a tool can apply only the changes between the version you are on and the one you are moving to. When a skill is unclear or a case is missing, the installed pota source is the ground truth.
These docs were written by a human, then rewritten, expanded, and reviewed with AI. Some examples are the original author's; others were generated, adapted, and verified during that pass. Every page is checked against the pota source — the source is the ground truth, and where the docs and the code disagree, the code wins. If you spot a mismatch, please open an issue.
pota was written by a human, from scratch — the reactive core, the renderer, the Babel preset, and the components are hand-built, with the lineage to Solid and dom-expressions described on the thanks page. More recently, AI has been used to iterate on and improve parts of it: tightening internals, expanding tests, and this documentation. The design and the decisions remain the author's.