GistUI is built to be installed by a coding agent as easily as by a person. There are three ways to give an agent what it needs.
The agent skill
@gistui/catalog ships a skill that matches the installed version:
SKILL.md: how to add GistUI to an app in every framework, with the Vercel AI SDK, tools, forms and theming.reference.md: the language itself, generated from the catalog. Every component, prop, enum and flag.
After npm install @gistui/catalog, tell your agent:
Read node_modules/@gistui/catalog/skills/gistui/SKILL.md and add GistUI to this app.
To keep it as a project skill, copy the folder:
cp -r node_modules/@gistui/catalog/skills/gistui .claude/skills/gistuiThese docs, as Markdown
Every page here has a Markdown version. Add index.md to a page’s address, or use the Copy page button at the top of a page.
| Address | What it is |
|---|---|
/llms.txt |
An index of every page with a one-line summary |
/llms-full.txt |
The whole documentation as one text file |
/<page>/index.md |
One page as clean Markdown |
/<section>/llms.txt |
The index of one section, for example /components/llms.txt |
Read https://gistui.com/docs/llms.txt and use it to answer questions about GistUI.
Writing GistUI Lang by hand
When an agent writes a program itself (a fixture, a test, an example), have it check its work:
import { library } from "@gistui/catalog";
import { parse } from "@gistui/core";
const r = parse(source, library);
// r.valid.strict: no errors or warnings at all
// r.valid.lenient: every problem was repaired deterministically
// r.errors: [{ code, stmtId, line, message, hint }]The rules that matter most are in Syntax; the exact text a model receives is in The full reference.