Free tool
Can ChatGPT read your page? Free AI page check
Last updated
Enter the URL of one page. We fetch its HTML once and read it without running any JavaScript, which is how the AI crawlers behind ChatGPT, Claude and Perplexity read it. You see what they get: the words in the HTML, the page cut into chunks at its headings, and what the audit's page-level rules find.
One request for the page's HTML, the way a crawler without JavaScript gets it. Nothing is stored.
What the check measures
- What a crawler without JavaScript receives. The HTTP status, the URL after redirects, the words of text in the raw HTML and the number of scripts. Almost no text with at least one script is a JavaScript shell: the content only exists after the browser runs the scripts, and AI crawlers do not run them.
- The page in chunks. The heading outline with the words in each section. Answer engines retrieve and quote passages, and split pages into passages at headings, so a long run without a heading is one large passage. Text is counted wherever it sits, including in plain
divelements, and counted once. - The audit's page-level findings. The title and meta description against the audit's bounds, the canonical, noindex in the robots meta tag or the X-Robots-Tag header, modified-date signals, statistics and quotations, and the structured data types present. Each finding says where it is, what was measured and why it matters, in the same words the full audit uses.
The rules it runs
These are the full audit's own rules, run on this one page, so a finding here is the finding the audit would file for it.
- Missing title
- Page has no <title> or the title is blank.
- Title outside head
- The page's <title> is in <body>, not <head> (streamed metadata).
- Title length
- Title is shorter than 20 or longer than 60 characters.
- Missing meta description
- Page has no meta description.
- Meta description length
- Meta description is shorter than 50 or longer than 160 characters.
- Missing h1
- Page has no H1.
- Multiple h1
- Page has more than one H1.
- Thin content
- Page has fewer than 150 words of body text.
- Http error
- Page returned a 4xx or 5xx status.
- Noindex
- Page is marked noindex (meta robots or X-Robots-Tag).
- Missing canonical
- Page has no canonical link.
- Canonical mismatch
- Canonical points to a different URL than the page itself.
- Client side rendered
- Page body is nearly empty without JavaScript; AI crawlers do not execute JS.
- Long section
- A section runs more than 250 words without a heading; engines chunk by heading.
- Missing date modified
- No modified-date signal (article:modified_time, dateModified) on the page.
- Stale content
- Content was last modified more than a year ago.
- No evidence
- Long page with no statistics or quotations; cited passages carry numbers and quotes.
What one page cannot show
Some of the audit's rules need the whole site, or something only the site's owner can declare. They are not run here, and the free audit covers them.
- Fetch failed
- Needs a URL the crawl found and could not read; here a failed request is reported as it happened instead.
- Broken internal link
- Needs the status of every page this one links to, which means crawling them.
- Orphan page
- Needs every page of the site, to know whether any of them links here.
- Slow page
- One request from a free tool is not a speed measurement; the audit times every page it crawls.
- Brand absent
- Needs the brand name, which is declared on a project in the audit.
- Missing sitemap
- A site-level check of /sitemap.xml and robots.txt.
- Ai search crawlers blocked
- A site-level check of robots.txt.
- Ai training crawlers blocked
- A site-level check of robots.txt.
- Gemini grounding blocked
- A site-level check of robots.txt.
- Content signals restrict ai
- A site-level check of robots.txt.
Whether AI crawlers may fetch the site at all, by robots.txt or at the CDN, is what the ai crawler checker checks.
Limits of the measurement
- The page is fetched once, from our server, with an ordinary browser user agent. A CDN that treats AI crawlers differently from browsers can serve them something else; the ai crawler checker compares the two.
- The first 2 MB of HTML are read. A longer page is measured on that part and the result says so.
- The request has 10 seconds and follows up to 5 redirects. Addresses that are not public websites are refused.
- A response that is not HTML, such as a PDF or an image, is reported as such.
- Nothing is stored. The result goes back to you and nowhere else.
Why it matters for AI visibility
An answer engine can only cite what its crawler could read, and it cites a passage, not a page. A page whose text arrives only through JavaScript is empty to the crawler. A page whose text is one long run under a single heading is one oversized passage; among cited pages, the most-cited quarter has 12.5 times the heading density of the least-cited quarter.
This is one page. The free audit reads up to 25 pages of the site this way, adds the site-level checks, and hands every finding with its evidence to your coding agent.