A markdown-based document standard for providing LLM guidance on websites
Core Idea: LLMs.txt is an emerging standard that provides links and guidance for Large Language Models (LLMs) related to a website, functioning like a table of contents that helps models discover relevant documentation.
Key Elements
- Purpose: Provides a structured reference of URLs and descriptions that LLMs can use to find relevant documentation
- Format: Simple markdown file with organized sections
- Function: Acts as a table of contents that models can scan to determine which specific documents to retrieve
- Location: Typically hosted at a predictable location on a website
Structure
- Organized in sections with headings
- Contains links to documentation resources
- Includes descriptions explaining what each link covers
- Categorizes documentation (e.g., how-to guides, conceptual docs, use cases)
Implementation
- Simple to create and maintain
- Can be generated by documentation systems
- Many libraries are building LLMs.txt files automatically
- Requires minimal setup compared to complex indexing systems
Connections
- Related Concepts: Context Loading Approaches for LLMs (provides context without requiring full indexing), Model Context Protocol (MCP) (helps connect tools that read LLMs.txt)
- Broader Context: Document Reference Management for LLMs (one approach to document management)
- Applications: Using LLMs.txt with Development Tools (practical implementation)
- Components: Tool Calling with LLMs (mechanism for retrieving documents referenced in LLMs.txt)
References
- LangGraph documentation and implementation examples
- Anthropic Claude, Cursor, and Windsurf integration examples
#llm #documentation #standard #context-management
Connections:
Sources: