All Tips
November 11, 2026
AI Search Visibility

llms.txt Is a Map for Agents, Not a Ranking Signal

llms.txt is a Markdown file that gives AI agents a short, curated overview of a site. Google Search ignores it and no crawler documentation promises to read it, so generate it from your content or skip it.

llms.txt is often sold as "robots.txt for AI" or as a way to rank in AI answers. It is neither. Here is what the proposal says, who has committed to reading it, and how the one on this site is built. Sources read on November 11, 2026.

What it is. llms.txt is a proposal by Jeremy Howard, published on September 3, 2024 and revised on August 10, 2026: "a proposal to standardise on using an /llms.txt file to provide information to help agents use a website". It is a Markdown file, at the site root or under a subpath such as /docs/llms.txt. The proposal separates it from the two files it gets compared to: robots.txt says what access is acceptable, a sitemap lists every page for search engines, and llms.txt "offers a curated overview". Its authors describe it as something an agent reads on demand, while helping a user, more than as training material.

The format. Only the first line is required:

# Site or Project Name

> One or two sentences saying what the site is.

## Docs

- [Page title](https://www.example.com/page): what the page covers

## Optional

- [Changelog](https://www.example.com/changelog)

An H1 with the name, a blockquote summary, then H2 sections holding lists of links, each link optionally followed by a colon and a note. A section named "Optional" holds links an agent can skip when it needs a shorter context. The proposal also suggests serving a clean Markdown version of a page at the same URL with .md appended.

Who reads it. The proposal names no AI product that does. Google is explicit: its AI optimization guide says Google Search does not use these files, and a note added on June 15, 2026 says keeping one "will neither harm nor help" your visibility or rankings. The crawler pages of OpenAI, Anthropic and Perplexity do not say their crawlers read a site's llms.txt either.

What several vendors do is publish one for their own documentation: on November 11, 2026, the developer docs of Anthropic (platform.claude.com), OpenAI, Perplexity, Cloudflare and Webflow each served an llms.txt. That is where the file is visibly in use today: documentation sites. A marketing site is a weaker case.

How mine is generated. The llms.txt of this site is not written by hand. A build step joins a template (the summary, services, page map and contact sections) with lists made from the content itself: every published case study in the site's order, then every post and tip, newest first, each with its absolute URL. A new article appears in the file at the next build, and an unpublished case study never does. Today the file is 125 lines and under 7 KB.

If you publish one, my checklist. The proposal requires only the H1; the rest is my advice.

  1. Serve it as plain text at /llms.txt. Some sites answer that URL with a styled HTML page, which is not an llms.txt.
  2. Start with a single H1, then a > summary line.
  3. Use absolute links in Markdown form, with a short note where the title is not enough.
  4. Curate. A list of every URL is a sitemap; pick the pages that explain the business.
  5. Generate it from your content, or put its update on the publishing checklist. A stale file describes a site that no longer exists.

I publish one because, generated this way, it costs nothing to keep current, not because any engine has promised to read it. If yours would be maintained by hand and forgotten, skipping it loses you nothing in Google.

My free SEO audit notes whether a site has an llms.txt without counting its absence against you, and warns when a published one is missing its title, summary or links.

Want This Level of Care on Your Site?

Start with a free audit of your site, checked by me. Or book thirty minutes with your project on screen: you leave with a scoped plan and a number, whether or not we work together.

Grégoire Geoffroy
Grégoire Geoffroy
Webflow Partner
Free audit
Report link within 3 business days
Call
30 min · Google Meet
Time zone
Europe/Paris
Reply
Within one business day