Privacy
The checker stores what it needs to show a report at its own address and to keep the service within its daily limits. This page lists all of it. The site sets no cookies and loads no analytics.
What we store when you check a page
| Data | Why | Kept for |
|---|---|---|
| The scanned address | The report lives at an address made from it. A new check of the same address within an hour shows the report already made. | Until the report is removed |
| The report, with samples of failing elements | So the report opens again at its address and can be shared. Each sample is the HTML of one failing element, as the page served it. | Until the report is removed |
| The page evidence | What the checks read from the page and the browser, so a stored scan can be checked again under newer checks. | 30 days |
| Your IP address, for the daily limit | Each address may check one site a limited number of times a day. For IPv6 we keep only the first 64 bits. | About 24 hours: removed by the first check started after 24 hours have passed |
The scanner also writes operational logs to Cloudflare, which keeps them for a short period it sets.
What we send to a model
When a report has a “Judged by a model” section, parts of the scanned page went to Cloudflare Clef (@cf/cloudflare/clef) on Cloudflare Workers AI: link texts, headings and labels, text alternatives of images, the start of the visible text, error messages of a form, and the text of an accessibility statement. Its answers are stored with the report.
Removing a report
To remove a report and its evidence, open an issue with the report's address. Issues are public, so leave out anything personal. Open one on GitHub issues.