llms.txt, and whether you need one.

We serve an llms.txt file on this site and we test that it stays accurate. We still cannot tell you it changes an outcome. Here is the honest version, including the part most articles leave out.

What the file is

An llms.txt file is a plain-text summary of your site, written in Markdown and served at the root. Markdown is the lightweight formatting syntax used for headings, links and lists. The idea is that a model reading your domain gets an accurate description instead of guessing from your navigation.

A conventional file states what the site is, who operates it, and which pages matter, with one line of description each. Ours does exactly that, and a test fails the build if a page enters our sitemap without being described there.

What it is not

It is routinely confused with robots.txt, and the two are unrelated.

	robots.txt	llms.txt
Purpose	Access control	Description
Status	Long-established	Proposed convention
Honoured by major crawlers	Yes	Not confirmed
Blocks anything	Yes	No

The part most articles skip

No major AI provider has publicly committed to reading llms.txt as a retrieval input. Anyone telling you it is required is ahead of the evidence. We ship one because it is cheap, not because we measured a gain.

We will also say the unhelpful-sounding thing. None of our ten scoring modules check for the file, so writing one will not move your grade here. These are the things that actually carry the AI-search half of the score, which totals 45%.

Entity Clarity, 12%. Whether a model can tell who you are and what you sell.
Citability, 12%. Whether there is a passage worth quoting.
Topical Authority, 8%. Whether you cover the subject rather than mention it.
AI Accessibility, 8%. Whether the page reads without running JavaScript.
Trust Signals, 5%. Whether an author, a date and a source are present.

What we would do first

If you have one hour, spend it on the page a buyer actually lands on rather than on this file. Add a plain definition of what you do. Add a table. Add three real questions with real answers.

Then write the llms.txt file, because an hour later it costs you nothing and it might matter later. The order is the whole point.

Common questions.

Do the major AI companies actually read llms.txt?

No major provider has publicly committed to reading it as a ranking or retrieval input. Treat it as a proposed convention rather than a supported standard. We serve one because it costs almost nothing, not because we can prove it changes an outcome.

Is llms.txt the same as robots.txt?

No. robots.txt is an access-control file that tells a crawler which paths it may fetch, and it is honoured by every major crawler. An llms.txt file grants no permissions and blocks nothing. It is a summary written for a reader, not a rule written for a machine.

Will an llms.txt file improve my report card score?

Not directly, and we would rather say so. None of our ten scoring modules look for the file. The things that do move your AI-search score are entity clarity, quotable content, readable HTML and trust signals, which are page-level properties.

If it is unproven, why write one at all?

Because the cost is an hour and the downside is nothing. It also forces a useful exercise: stating plainly what your site is, who runs it, and which pages matter. Most sites cannot answer those three questions in one place.

The free scan reads one page and scores the five AI-search modules above, so you can see which of them is actually holding you back. Paid teardowns start at $79 and crawl more of the site.

Run a free scan →