Website cloaking checker: paste a URL, get the page as each visitor type sees it, and see where the views differ. Cloaking means showing crawlers or reviewers one page and visitors another. Background: website cloaking.
What we check
Five requests, one per view. Desktop Chrome is the baseline; every other view is compared to it.
| view | user agent | referrer |
|---|---|---|
| Desktop Chrome | Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/142.0.0.0 Safari/537.36 |
none |
| iPhone Safari | Mozilla/5.0 (iPhone; CPU iPhone OS 18_5 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/18.5 Mobile/15E148 Safari/604.1 |
none |
| Googlebot | Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; Googlebot/2.1; +http://www.google.com/bot.html) Chrome/142.0.0.0 Safari/537.36 |
none |
| Facebook crawler | facebookexternalhit/1.1 (+http://www.facebook.com/externalhit_uatext.php) |
none |
| Ad click | Desktop Chrome | https://www.google.com/, URL plus gclid=…&utm_source=google&utm_medium=cpc |
Every view sends the same Accept and Accept-Language: en-US, so only the user agent and the referrer change. Background: user-agent cloaking, referrer cloaking.
What we compare
Each view against Desktop Chrome:
| field | a difference counts when |
|---|---|
| Final URL and redirect chain | the view ends on another address (www, m. and query are ignored) |
| Status | the view is refused (HTTP 4xx/5xx or no answer) |
| Visible text | under 60% alike (3-word overlap); 60-90% is noted only |
| Title | differs, and the text differs too |
| Link domains | the sets overlap under 50% |
| Form, checkout | present in one view and not in the other |
| Script signals | found in a view and not in Chrome |
Script signals: JavaScript is read, not run. We look for meta refresh, location redirects, document.write of big or decoded blocks, encoded eval, bot or referrer checks, ad click ids, and the names of known cloaker tools. A signal is a lead: scripts can change the page after load, and we do not see that.
Verdict: same, cloaked (a view got another page), blocked (a view was refused, or Chrome got an empty page) or geo-blocked (HTTP 451 or a “not in your country” page).
Limits
- One datacenter location (Cloudflare). No country, residential or mobile IPs, no real browser, no JavaScript run: contact us for a deep scan.
- Public http and https pages only, ports 80 and 443. No logins, no private or internal addresses.
- 8 seconds and 2 MB per view, up to 5 redirects.
- 5 checks per visitor per day, 6 per site per hour.
- Results are not stored. We keep only counters (hashed visitor, site, day).
- A refused Googlebot or Facebook view is often bot protection: our request carries their user agent, not their IP.
What we do not check yet
- Country views. Residential and mobile IPs. Both are a deep scan: contact us.
- Real Googlebot IPs. Google lets site owners verify Googlebot by reverse DNS (Google docs). A site that does this sees our Googlebot user agent coming from a non-Google address. See IP-based cloaking.
- Anything behind a login.
Read the result
- Same: no difference found from this exit, with these views.
- A view differs: open the diff, then check it against what is allowed.
- Redirects differ: see redirect cloaking.
To reproduce by hand: check a site with curl.