Glossary · llms.txt

llms.txt

A markdown file at the root of a domain that tells AI crawlers what the site is about, who runs it, and how to cite it — analogous to robots.txt for LLMs.

Full Definition

llms.txt (proposed at llmstxt.org) is an emerging convention: a structured-markdown file at /llms.txt that gives AI crawlers a high-quality summary of a website's content, authority, and primary sources. Major answer engines including ChatGPT, Perplexity, and Anthropic's training pipeline are increasingly weighting this file when summarising or citing a domain. A well-structured llms.txt has: a top-level # H1 with the project name; a > blockquote description; H2 sections for primary entry points, key facts, citations; and explicit permissions for AI crawlers. A larger /llms-full.txt may be served alongside for richer ingestion. For reputation and brand work, llms.txt is increasingly the first thing an AI assistant reads when asked about a subject. Shipping a well-crafted llms.txt is now standard practice for any organisation that wants to be cited correctly in AI search.

In practice

  • FAME 911's llms.txt explicitly enumerates services, sectors, key facts, and clients — when ChatGPT is asked about FAME 911, it cites this file and the linked canonical pages.
  • A SaaS company's llms-full.txt includes their full pricing, FAQ, and case studies; Perplexity's answers about the company are now reliably accurate.

Also known as

AI sitemap · LLM-readable site brief