Free interactive tool

llms.txt Validator

Paste your llms.txt and the checker reports whether it follows the llmstxt.org file structure: an H1 title (the file format’s only required element), a blockquote summary, and one or more H2 sections whose list items are markdown links an agent can follow. It flags the mistakes that break parsing silently, such as a summary written as a plain paragraph instead of a blockquote, or a list entry that is a bare URL instead of a link. It runs in your browser and fetches nothing. It does not and cannot check whether any AI system reads, follows, or ranks the file — as our own llms.txt decision guide covers, Google says plainly that Search “doesn’t use” files like this one.

What the checker looks for

1  H1 on the first non-blank line: the title. The only required
     element. Missing it fails the check.
2  A blockquote ("> ") right after the title: the summary.
     A plain paragraph in that spot is flagged, not accepted.
3  Zero or more H2 sections. Each list item must parse as a
     markdown link "[name](url)": a bare URL is an error.
4  A section named "Optional" is read as the by-convention list
     of links an agent can skip.
5  H3+ headings, duplicate section titles, and empty sections
     are flagged as outside or against the spec's intent.

Source: the structure checked here is read from llmstxt.org (spec revision dated 2026-08-10, read 2026-09-26): an H1 title is the only required section; a blockquote summary, non-heading context sections, and H2 “file list” sections of markdown links follow it; an “Optional” H2 is a documented convention for links an agent can skip. This is a third-party convention, not a Google standard. Google’s own AI-optimization guide (updated 2026-07-10) states that Search “doesn’t use” files like this to appear in results, so a passing format check is not a ranking or citation signal of any kind — only a statement that the file itself is readable the way the spec describes.

Check your file

Start from:

Your llms.txt

Paste the file’s contents. It is checked in your browser as you type; nothing is sent anywhere.

Format check1 error

Name (H1)	Example Docs
Summary (blockquote)	None found
File-list sections	Guides (3), Optional (1)
Links an agent can follow	3
Size	449 characters, roughly 112 tokens (an estimate at four characters a token, not a model’s real tokenizer)

Findings, most serious first

Error (line 8): Line 8, in "Guides": a bare URL (https://example.com/docs/auth) is not a markdown link. Use "[name](https://example.com/docs/auth): notes" so an agent parsing links can find it.
Warning (line 3): The line "Example Docs is the help center for the Example billing API." looks like a summary, but it is a plain paragraph, not a blockquote. The spec's summary line starts with "> ".
Warning (line 11): H3 heading found. The spec defines only H1 (title) and H2 (file-list section headers); deeper headings are not part of the structure and a strict parser may not recognize this as a section.
Note: Content appears between file-list sections rather than only before the first H2. The spec allows non-heading sections before the file lists; content wedged between H2 sections is unusual but not explicitly forbidden.

Worked example: a file with three common mistakes

The starting example above is a hypothetical support-docs file with the three mistakes this checker is built to catch: the summary line after the H1 is a plain sentence, not a blockquote, so it is not read as the file’s summary; the “Authentication” row is a bare URL, not a markdown link, so an agent parsing links for file lists will not see it; and the “Advanced” section uses an H3, which is not part of the H1/H2 structure the spec defines. Fix the blockquote and the bare link, and the file passes with zero errors — the H3 issue stays a warning, because nothing in the spec bars extra markdown, it only means a strict parser may not read “Advanced” as a file-list section the way ## Guides is.