Reference

llms.txt

Standard

Machine-readable API reference you can paste directly into any LLM context window.

What is llms.txt

llms.txt is a proposed open standard — analogous to robots.txt — for helping LLMs understand a project. It lives at the root of a site, uses plain Markdown, and provides a short structured index with links to deeper content.

Atelier UI provides two files: a short index (llms.txt) and a comprehensive reference (llms-full.txt) containing every component's props, types, defaults, and framework-specific usage examples across Angular, React, and Vue.

The files

llms.txt ~2 KB

Short index: component list with one-line prop summaries and links to docs. Use this when your context window is limited or you just need component names and key props.

https://atelier.pieper.io/llms.txt

llms-full.txt ~12 KB

Full reference: every component with all props (name, type, default, description), framework-specific usage examples for Angular, React and Vue, design token system, and accessibility notes.

https://atelier.pieper.io/llms-full.txt

How to use it

Paste into your LLM context

Copy the URL of llms-full.txt and ask your LLM to fetch it, or paste the file contents directly. The file is plain text optimised for token efficiency — no HTML, no markup noise.

Here is the full API reference for the Atelier UI component library:

<paste contents of llms-full.txt here>

Using this library, build a settings page with a card containing
name (Input), email (Input), and notification preference (Toggle).

Reference via MCP

The Atelier UI MCP server provides structured access to component docs so your LLM can look up components on demand without consuming your entire context window upfront.

{
  "mcpServers": {
    "atelier-ui": {
      "url": "https://atelier.pieper.io/storybook-angular/mcp"
    }
  }
}

Add to CLAUDE.md or .cursorrules

For persistent project-level context, reference llms.txt from your project's AI instructions file so every session starts with component awareness.

# UI Components

This project uses Atelier. Full API reference:
https://atelier.pieper.io/llms-full.txt

Key import pattern:
import { LlmButton, LlmInput, LlmCard } from '@atelier-ui/react';

Why plain text

Token efficient

No HTML tags, no surrounding markup. Every token carries signal.

Framework-neutral

One file covers Angular, React, and Vue side by side for easy comparison.

Always fresh

Generated from source — reflects the current release, not outdated docs.

Cacheable

LLM tools can cache the file across sessions. Fetch once, reference many times.

Preview

A snapshot of what an LLM sees when you share the reference file:

# Atelier UI — Full API Reference

> Complete component API for LLM consumption. 24 accessible components
> for Angular, React, and Vue with consistent prop naming.

---

### LlmButton

  Props:
    variant   'primary' | 'secondary' | 'outline' | 'danger'   'primary'   Visual style
    size      'sm' | 'md' | 'lg'                     'md'        Size
    disabled  boolean                                 false       Prevents interaction
    loading   boolean                                 false       Shows spinner

  Angular:  <llm-button variant="primary" [loading]="isSaving">Save</llm-button>
  React:    <LlmButton variant="primary" loading={isSaving}>Save</LlmButton>
  Vue:      <LlmButton variant="primary" :loading="isSaving">Save</LlmButton>

---

### LlmCombobox

  Props:
    value        string               ''    Selected option value
    options      LlmComboboxOption[]  []    Array of {value, label, disabled?}
    placeholder  string               ''    Input placeholder
    disabled     boolean              false Disables the combobox
    invalid      boolean              false Error styling

  React:
    const options = [
      { value: 'us', label: 'United States' },
      { value: 'ca', label: 'Canada' },
    ];
    <LlmCombobox value={country} onValueChange={setCountry}
      options={options} placeholder="Search..." />

...and 22 more components