The framework-agnostic AI-chat library
Chat, spoken aside.
Vanilla web components at the core, thin wrappers for React, Vue, Svelte and Angular — and zero runtime dependencies. Point it at any model through a transport; the key stays wherever you want it.
$ npm i @aparte/core assistant · try it
i.Zero runtime dependencies ii.Five ways to mount it iii.Streams by default iv.Your key, your server
Same component, (every framework)
It’s one web component underneath, so the rendered chat is identical everywhere. What changes is the two lines you write to mount it.