FREE TOOL · DETERMINISTIC · BOUNDED FETCH

See whether your llms.txt is a briefing or a stub.

We read the live file and check structure, notes, and a few advertised links. No crawler is bound by llms.txt — this is a hygiene check, not a prediction of whether AI will cite you.

No account. One bounded check — never the full crawl or a site-wide score.

WHAT YOU GET

ILLUSTRATIVE

File class: briefing — hygiene, not a citation score

This file looks like a briefing.

CHECKS

h1 · present
note_coverage · present
coverage · unknown

What this check actually measures

llmstxt.org requires exactly one thing: an H1. Everything else is optional, so “has llms.txt” discriminates almost nothing. This tool fetches the live file, checks whether it is structured markdown with notes on its links, and probes a few advertised same-origin URLs. The headline is a file class — briefing, stub, missing, unreadable, or unknown — not a score, and not a prediction of whether any engine will cite you.

The failure modes, and what each one means

We could not read the file: A timeout, a 403, or a 5xx is an absence of information. That is not “you have no llms.txt”, and it is not a stub. Fix: /llms.txt should return 200 as text, or 404 if you genuinely do not have one.
The file is HTML, empty, or a stub: A 200 that is an error page, an SPA shell, or an H1 with no sections is not a briefing. Fix: serve markdown, add a summary, and give each link a note about what it is.
Advertised links we did not fetch: We probe at most eight same-origin URLs. Offsite links are counted, not fetched. Anything past the cap is unknown, never a clean pass. Coverage against a sitemap is not measured here.

Why it matters — and what it does not claim

No crawler is bound by llms.txt. A briefing is hygiene: it tells an agent what you think the site is. It is not Reach, not retrieval, and not a ranking signal. Treat a stub as a file that teaches nothing; treat a missing file as a missing file; do not read either as “AI will not cite you”.

What the full product adds

This tool grades one URL, once, with no site inventory. The TrustGrowth audit runs the same grader against the pages it already knows and files hygiene issues when the file is a stub or unusable. This check does not change a published TrustGrowth Score.

Convinced? It takes one URL.

TRY ANOTHER CHECK

Check my schema → Build JSON-LD → Build sameAs JSON-LD → Preview my listing → Inspect my canonical → Check crawler access → Create my llms.txt → Compose my robots.txt → See the AI read → Check trust signals → Ask three engines → Convert this page →

Browse all free tools The full version lives in AI Visibility →

Frequently asked questions

What does this tool actually check?

The live /llms.txt on the domain you name: whether it can be read, whether it is markdown with an H1 and sections, whether links carry notes, and whether a few advertised same-origin URLs resolve. It does not crawl the rest of the site, and it does not consult a sitemap.

If the grade is good, will AI cite me?

No. No crawler is bound by llms.txt. A briefing is hygiene — a file that tells an agent what you think the site is. It is not Reach, not retrieval, and not a ranking or citation score.

What is the difference between a stub and a missing file?

Missing means a usable 404: there is no llms.txt. A stub means a file was found and it teaches almost nothing — an H1 only, fewer than three unique URLs, or every link pointing at the same URL. Those are different findings, and an unreadable response is a third state, not either of them.

You could not read my file. Does that mean I do not have one?

No. A timeout, a 403, or a 5xx is an absence of information. The tool reports that as unknown, never as missing, and never as a stub.

Why are some advertised links unknown?

The free tool probes at most eight same-origin URLs. Offsite links are counted, not fetched. Anything past the cap, and any inventory coverage against your sitemap, stays unknown rather than a measured zero. Absence of llms-full.txt is not a defect.