LLM Documentation Files

Sugarcoat provides llm.txt and llm-full.txt files that offer comprehensive documentation in a format optimized for AI assistants. These files follow the emerging llms.txt standard↗ proposed for AI-friendly documentation.

What are llm.txt Files?#

The llm.txt standard provides a way to expose documentation in a format that AI assistants can easily consume. Unlike MCP which provides real-time interaction, llm.txt files are static documentation that can be:

• Included in AI assistant context windows
• Referenced by AI crawlers
• Used with AI tools that don't support MCP

Hosted Documentation#

Sugarcoat hosts LLM documentation files at:

File Types#

llm.txt#

A concise index of components with links to documentation. This file is small (~5-15 KB) and suitable for inclusion in AI context when you need a quick reference.

Contents:

• Quick start guide
• Links to all components grouped by category
• Design system overview
• Resource links

llm-full.txt#

Complete inline documentation (~50-200 KB) including:

• All component props and types
• Usage examples with code
• Source code snippets
• Design token reference
• CLI command reference

This file provides AI assistants with everything they need without following links.

Using with AI Assistants#

ChatGPT / Claude Web#

When using AI assistants without MCP support:

1. Reference the URL directly:

   "Please read https://sugarcoatui.com/llm-full.txt↗ for Sugarcoat component documentation"

2. Copy-paste content: Copy relevant sections into your conversation

3. Upload as file: Some AI assistants allow file uploads - download and upload llm-full.txt

Custom GPTs / Claude Projects#

When creating custom AI assistants:

1. Add https://sugarcoatui.com/llm-full.txt as a knowledge source
2. Or include the file content in the system prompt
3. The hosted files are always up-to-date with the latest components

IDE Extensions#

Some IDE AI extensions allow custom documentation sources:

1. Point to https://sugarcoatui.com/llm.txt or llm-full.txt
2. The AI will have context about Sugarcoat components

Comparison with MCP#

Recommendation: Use MCP for supported AI assistants (Claude Code, Cursor, VS Code) and llm.txt as a fallback for other tools.

File Size Considerations#

For AI assistants with limited context windows, prefer llm.txt. For those with larger contexts (like Claude), llm-full.txt provides better assistance.