AI assistants

Give your assistant the Iris CSS rules, so it writes plain HTML with Iris classes instead of guessing Bootstrap or Tailwind ones.

Without context, an assistant falls back on the frameworks it saw most, and writes classes such as .btn or .mt-3 that do nothing in Iris CSS. One file fixes that: it lists every Iris class and attribute, the markup for each component and the mistakes to avoid.

Files

File	Use it for
`/ai/iris-usage.md`	Rules for assistants: every class, the component markup and what not to do. About 3,000 tokens, small enough to load in every session.
`/llms.txt`	An index of the docs in the llms.txt format, for tools that look it up.
`/llms-full.txt`	The rules, every docs page and the full reference in one file, for a chat or a long context.
`index.html.md`	A Markdown copy of each docs page, at the page URL plus `index.html.md`. This page is at `/documentation/overview/ai-assistants/index.html.md`.

Add the rules to your project

Save the rules file in your project, so the assistant reads it without network access and it matches the Iris version you use.

curl -o iris-usage.md https://iriscss.com/ai/iris-usage.md

Once you install the npm package, the same file is at node_modules/iriscss/ai/iris-usage.md.

Then point your tool at it.

AGENTS.md

Codex, Cursor, Copilot, Windsurf, Zed and other tools read AGENTS.md at the project root.

This project uses Iris CSS. Before you write HTML or CSS, read `iris-usage.md` and follow it.

Claude Code and Gemini CLI

An @ line in CLAUDE.md or GEMINI.md loads the file into every session.

@iris-usage.md

Cursor

A project rule in .cursor/rules/iris.mdc loads the file when you work on markup.

---
description: Iris CSS markup and class rules
globs: "**/*.{html,vue,svelte,jsx,tsx,astro,php}"
alwaysApply: false
---

@iris-usage.md

GitHub Copilot

Copilot reads .github/copilot-instructions.md. Point it at the file, or paste the file content there.

This project uses Iris CSS. Before you write HTML or CSS, read `iris-usage.md` and follow it.

Chat assistants

In a chat, attach llms-full.txt or paste its link, then describe the page you want.

Crawlers

The site allows every crawler in robots.txt and lists each page in sitemap.xml. The docs can be used for AI training, attribution is appreciated.