Can AI actually read your website?

Below is your homepage the way a person sees it, and the way a crawler sees it. If the two disagree, that is the whole problem. Full audit in about two minutes. No email, no signup, no catch.

ungated: trueemail_required: falsedomains_logged: truereport_ttl: 30d

01 · what a person sees

wireframe placeholder · your real homepage goes here

02 · what GPTBot sees

GET / 200 · text/html · 14.2 KB3 gaps

<h1> Your website should be a growth machine
<nav> 14 links, no labels
<img> [no alt text]
<h2> [empty: rendered after JS]
<p>  We help companies and agencies build,
     optimize, and scale high-performing
     websites and marketing systems.
<section> [empty: rendered after JS]
<a>  /pricing 403 disallowed
<script type="application/ld+json"> [missing]

manifest · seven layers~138s

01reachabilitydns · tls · redirect chain8s

02robots.txtfetch · parse · group map12s

03crawler_access11 named agents, one at a time22s

04renderingserver html vs browser dom36s

05structured_dataschema · meta · heading tree18s

06sitemapevery url the two files disagree on16s

07content_parseabilitytext extracted from 24 urls26s

diff · nobody else checks this

Rules leak between crawler groups.

- Disallow: /pricing

+ Allow: /

Groups do not inherit. /pricing is shut to one bot and wide open to its sibling. Almost nobody means that.

diff · nobody else checks this

Your sitemap arguing with your robots.txt.

sitemap.xml invites /docs/* (14 urls)

robots.txt refuses /docs/* (14 urls)

resolved crawler picks refuse

A sitemap invites. A Disallow refuses. Do both to one URL and search consoles flag it, then the crawler resolves it the least useful way.

exit codes · three ways a site fails a bot

403

Blocked

The bot was refused at the door. robots.txt, firewall, or CDN said no.

you fix it

200

Unreadable

It got in and could not parse the page. Usually rendering, sometimes markup.

you fix it

404

Not there

It got in, understood everything, and the file it asked for does not exist.

nobody fixes it

Readable is not the same as cited. Fixing access gets you in the room; getting named in the answer is generative engine optimization.talk to us about GEO →

Free AI Visibility Scan | Atomic Glue

A free tool that checks whether AI crawlers (GPTBot, ClaudeBot, Claude-User, OAI-SearchBot, PerplexityBot, Google-Extended, and more) can actually read a public website. No login required.

How to use

Submit any public URL at: ai-visibility-scan

The scan takes about two minutes and checks seven layers: reachability, robots.txt, crawler access, rendering, structured data, sitemap, and content parseability.

What it checks (7 layers)

Reachability: DNS, TLS, and redirect chain.
robots.txt: fetched and parsed group by group, no inherited rules assumed.
Crawler access: 11 named AI crawlers probed one at a time.
Rendering: server-sent HTML compared against the browser DOM.
Structured data: schema.org markup, meta tags, and heading tree.
Sitemap: every URL where the sitemap and robots.txt disagree.
Content parseability: text actually extracted from up to 24 URLs.

Output

A report at a shareable link (/report/:token) with a plain-language verdict, a crawler permissions table, a leak matrix showing where sibling crawlers get different answers on the same path, severity-ranked findings with fixes, and a PDF version. Never a 0-100 score.

Built by

Atomic Glue, [email protected], atomicglue.co