Free tool · no signup

llms.txt: validate it, or draft it.

llms.txt is a short Markdown map of your site for AI readers — a title, a one-paragraph summary, and sections of links with a note on each. Enter your domain: if you have one, we check it against the format and try its links; if you don't, we draft one from your sitemap and page titles that you can copy straight to /llms.txt.

Reads /llms.txt, /llms-full.txt, /sitemap.xml and up to 24 of your pages. Cached for 10 minutes. Nothing stored. Does llms.txt actually help? →

About this check

What the format looks like

Per llmstxt.org: an # H1 title first, an optional > blockquote summary, optional free text, then ## Sections whose items are - [Page title](url): one-line note. A ## Optional section marks pages an AI reader can skip. llms-full.txt is the long-form companion with the actual content inlined.

What we validate

Plain text (not an HTML page), H1 first, summary present, H2 sections present, every list link well-formed, notes on links, links on your own domain, file size, and whether a sample of up to eight links answers 200. A failed check is a suggestion, not a verdict — the format is deliberately loose.

How the draft is built

From your sitemap.xml (one level of sitemap index followed), homepage first, shallow paths first, up to 24 pages. Each page contributes its <title> (site name stripped) and meta description as the note; pages are grouped by their first path segment. It is a starting point — trim it to the pages you actually want an AI to read and rewrite the notes in your own words.

Does llms.txt make AI recommend me?

No engine has confirmed it as a ranking or citation signal, and crawler-log studies show the file is rarely fetched so far. It is cheap, harmless and occasionally read by agents — worth ten minutes, not a strategy. What moves recommendations is whether the engines can read you and what they find; the free Reflexa check shows that side.

A map helps. Being the answer is the goal.

The free check runs your real buyer questions across ChatGPT, Claude and Perplexity — mentioned, recommended and cited, per engine, with the answers as evidence.

Run the free check →