llms.txt Standard Specification

llms.txt Specification: Structuring Sites for Generative Inference

Large language models require clean data structures. The llms.txt standard organizes web content specifically for LLM context windows.

1. Document Structure Rules

• H1 Title: Declares the name of the system or site.
• Blockquote: Contains a 1-2 sentence high-level summary.
• H2 Categories: Grouping documents using bulleted link formats: - [Title](URL): Description.
• Optional H2: Declared as ## Optional to isolate low-priority links.

2. The .md URL Extension

Each user-facing HTML page should serve its exact contents in markdown by appending .md to the path, providing clean inputs to developer tools.