AI-Optimized Documentation

Machine-readable documentation for AI assistants, coding copilots, and LLM-powered tools.

What is llms.txt?

The llms.txt convention provides AI assistants with structured, plain-text documentation optimized for context windows. This helps tools like Claude, ChatGPT, GitHub Copilot, and Cursor understand how to use @sveltopia/colors correctly.

Available Files

llms.txt

Quick overview with core concepts, installation, and links to detailed documentation.

~1KB — Best for quick context

llms-full.txt

Complete reference with all APIs, export formats, CLI usage, and accessibility guidelines.

~8KB — Full documentation

How to Use

With Claude, ChatGPT, or similar

Paste the contents of llms-full.txt into your conversation, or ask the AI to fetch it:

"Read https://colors.sveltopia.dev/llms-full.txt and help me generate a brand palette"

With Cursor or VS Code + Copilot

Add to your project's AI context or reference the URL in your prompt:

@https://colors.sveltopia.dev/llms-full.txt

Download for offline use

Save locally and include in your project's documentation context:

llms.txt llms-full.txt

What's Included

The full documentation covers:

Installation & CLI — Generate and dev commands, interactive mode, config files
12-Step Scale System — Radix step semantics, Tailwind mapping, safe color pairings
Export Formats — CSS, JSON, Tailwind v4, Tailwind v3, shadcn, Radix, Panda CSS
Programmatic API — generatePalette, analyzeBrandColors, all export functions
Accessibility — APCA validation, contrast thresholds, auto-remediation
Key Types — Palette, Scale, TuningProfile, ContrastReport
Common Patterns — Generate + export workflows, CI validation, brand analysis

About the Convention

The llms.txt convention was popularized by the AI development community as a way to provide documentation that's optimized for LLM context windows. Unlike HTML documentation, these plain-text files:

Use minimal tokens (no HTML/CSS overhead)
Include complete code examples
Follow a consistent, parseable structure
Can be easily fetched and included in AI prompts