llms.txt checker

Does your site really serve an llms.txt?

We checked all 500 companies in the S&P 500. Plenty of sites answer 200 at /llms.txt and serve a marketing page, a redirect, or nothing at all. This checks what is actually there.

We request https://your-domain/llms.txt and one randomized control path. Nothing is stored.

This is a publishing check. It tells you what your server returns at one path. It does not tell you that an AI assistant reads the file, and it does not tell you anything about your search rankings.

Method

How the check works

Five steps, in this order. Each one exists because a real site fooled a simpler version of this check.

  1. Fetch the path

    Request https://your-domain/llms.txt, follow redirects, read up to one megabyte, and give up after fifteen seconds.

  2. Reject HTML

    Check the content type, then check the first bytes of the body. Some servers label a web page as plain text. A page is not a file.

  3. Watch where you land

    Compare the final path with the requested one. A hop from http to https that still ends at /llms.txt is fine. A hop to the homepage is not a file.

  4. Ask for something that cannot exist

    Request a randomized path on the same domain. If it returns the same body, the server answers 200 for everything and the file is a mirage.

  5. Keep unknowns unknown

    A 403, a timeout, or a server error is reported as inaccessible. Calling those absent would turn our inability to measure into a claim about you.

What the answers mean

Five answers, not two

Valid and invalid is the wrong pair of buckets. Here is what we report instead.

Title and links
A real text file with a Markdown H1 and at least one Markdown link. Only 54 of the 500 S&P 500 companies cleared this bar.
Detected, title only
A real text file with an H1 and no Markdown links. The proposal allows this. It just does not curate anything for a reader.
Nonstandard
A real text response with no Markdown H1. The file exists. The shape does not match the proposal.
No file
The server answered, and what came back was a redirect, an HTML page, a generic fallback, an empty body, or an honest 404. This is the most common outcome by a wide margin.
Could not check
A 403, a timeout, a network failure, or a server error. Your file may well exist. A bot filter in front of it is a finding of its own.

The llms.txt proposal makes the H1 the only required part. Link sections are optional. So a file without links is nonstandard by our threshold, not invalid by the spec.

The research

Where these rules came from

We ran this method across the S&P 500 before turning it into a tool.

Contact

Talk to us

Ask us anything about your AI visibility, or get a guided walkthrough for your team.

  • 01Current AI brand visibility
  • 02Accuracy and risk gaps
  • 03Guidance from a dedicated expert

Typically respond within 24 hours. No automated follow-ups.

By submitting, you agree to our Privacy Policy