llms.txt: what it is, who reads it, and a template

llms.txt is a Markdown file at the root of a site that summarises it for large language models: a title, a one-line description, and links to the pages that matter. It's a proposal from llmstxt.org, not a standard. In 2026 about one in ten sites has one, but server log studies found that 97% of those files receive no AI requests at all.

What goes in the file

The proposal at llmstxt.org asks for an H1 with the site name, a blockquote summary, then sections of links with one-line descriptions:

# Example Co

> Example Co makes scheduling software for dental clinics.

## Product

- [Features](https://example.com/features): what the software does
- [Pricing](https://example.com/pricing): plans and prices

## Docs

- [API reference](https://example.com/docs/api): REST API for bookings

Publish it at /llms.txt. A copy at /.well-known/llms.txt covers tools that look there. An optional /llms-full.txt carries the full text of your docs in one file.

Does anyone read it?

Adoption grew fast: the number of sites with llms.txt rose 8.8 times to May 2026 (Originality.ai, reported by ppc.land), and measured adoption ranges from 8.7% of the top 1,000 sites to 11.8% across 151.6 million live sites (Rankability, June 2026; Stackscan, 2026).

Usage didn't follow. A 2026 analysis reported by ppc.land found 97% of llms.txt files received zero requests from AI systems, and Stackscan found two fifths of files were auto-generated by Wix and GoDaddy.

So add one, because it takes an hour, but fix readable HTML and robots.txt first. That's why CrawlReadable gives llms.txt 4 points out of 100 and server-rendered content 40.

Common mistakes

Serving your HTML homepage at /llms.txt: many sites answer every unknown path with the homepage. We check that the file isn't HTML.

Listing every page: the point is a short, curated map. Link the 5 to 30 pages that explain what you do.

Letting it go stale: if your pricing page moved, the file now points AI systems at a 404.

Questions