llms.txt

The llms.txt proposal, published by Jeremy Howard on September 3, 2024 and last revised on August 10, 2026, suggests adding a /llms.txt Markdown file to websites to provide LLM-friendly content. The revised version also allows the file at other paths within a site. The idea is that a language model reading a site at answer time gets a short, clean overview instead of navigation, scripts and cookie banners.

The format is simple:

An H1 with the site's name. This is the only required section.

A short summary in a blockquote, followed by any extra detail.

H2 sections with lists of Markdown links to the pages that matter, each with a short note. By convention, a section called "Optional" holds secondary links that an agent can skip.

The proposal also recommends offering a clean Markdown version of important pages at the same URL with .md added.

Be careful with promises. Google's documentation on AI features in Search says you don't need to create new machine-readable files, AI text files or markup to appear in those features. Google's John Mueller wrote on Bluesky in June 2025 that no AI system currently used llms.txt, as reported by Search Engine Roundtable; that was his view at the time, not a measurement. We have not found an official statement from a major AI provider that it reads the file.

So llms.txt is cheap to publish and does no harm, but it is not a ranking lever. For a store, the things that are known to matter for AI shopping are the ones that make products readable everywhere: complete product structured data, crawlable pages, AI crawlers allowed in robots.txt, accurate prices and stock, and consistent brand information. See AI agent readiness.

1. Open /llms.txt on the domain. A real file is plain text or Markdown starting with an H1, not an HTML page or a redirect to the homepage.

2. Check that the links work and point to useful pages, such as category pages, shipping and returns, size guides and the about page.

3. Keep it in step with the site. An llms.txt that lists retired categories or old policies is worse than none.

4. Weigh it correctly. RivalSweeper's AI Readiness Score checks whether each site publishes a valid llms.txt, as one small item in its AI-specific optimization pillar, next to signals such as agentic checkout through UCP and MCP endpoints. It gives the file little weight because no major AI provider has said it uses it. The free AI Readiness check runs a lighter version on any store.