llms.txt is a Markdown file at the root of a domain (https://example.com/llms.txt) that provides AI engines with a curated index of the site's content, structured like a hybrid of robots.txt and sitemap.xml.
Proposed by Jeremy Howard (fast.ai) in 2024 and adopted opportunistically by Anthropic, Perplexity, and others, llms.txt is a community standard rather than a formal specification. The file uses Markdown headings to organize content: H1 = brand name, blockquote = brand summary, H2 sections for Products, Documentation, Pricing, etc., each with bulleted links.
A companion file, llms-full.txt, contains the concatenated long-form Markdown body of every public document on the site (typically 50–500KB). Engines fetch llms.txt for orientation and llms-full.txt for retrieval at answer time.
llms.txt does not affect classic Google SERP rankings. Its value is in AI search citation (Perplexity, ChatGPT Search, Claude) and AI Overview placements.