About the Sigila scanner

What visits your site when Sigila scans it, and how to keep it out if you prefer.

What it is

Sigila checks public web pages for accessibility failures against WCAG 2.1 AA. A scan opens the page in a real Chrome browser, waits, closes the cookie dialog the way a visitor would, scrolls through the page and runs automated checks. It never signs in, never submits a form, and does not try to get past bot protection: if your site shows it a challenge, the scan stops and reports that.

When it visits

How to recognise it

Every scan comes from the address 94.237.112.217 and adds this to the browser's User-Agent string:

SigilaScanner/1.0 (+https://getsigila.com/bot)

Sigila also has test visitors: a browser that moves through a site the way a person would (with a mouse, a phone, the keyboard or a screen reader) to see whether something can be found, such as a phone number or a price. A test visitor uses the site's own links and menus. In an online shop it also searches with the site's own search box and puts products in the cart. When the site's owner asked for the test, it goes on through checkout up to the payment step, choosing a delivery option and filling in only Sigila's own test details (Test Sigila or Testi Sigila, crashtest@getsigila.com); on any other site it stops at the checkout page. It never pays, chooses a payment method, confirms an order or signs in, and never types in card, bank, identity or password details. A cart or an unfinished order it leaves may remain in the shop's system. It comes from the same address and adds this instead:

SigilaPersona/1.0 (+https://getsigila.com/bot)

How to keep it out

Research scans obey your robots.txt. To exclude them, add:

User-agent: SigilaScanner
Disallow: /

Our own test visits check it too. To exclude them, add:

User-agent: SigilaPersona
Disallow: /

To exclude your site from every scan, including ones visitors ask for, write to us through the support form (topic "Do not scan my site") from an email address at your site's own domain. We then block the site and its subdomains.

A site kept out of research is counted in our publications as "not scanned at the operator's request" or "not scanned (robots.txt)", never dropped silently, so that an exclusion cannot hide a result. Research scans never use free scans or customers' scans, and they follow fixed limits: public pages only, no sign-in, no forms (a test visit in a shop uses only the search box and the cart), no security testing, nothing that gets past a block. The privacy notice describes the research scans and their legal basis.

What the scanner changes on your page

Almost nothing, and it says so here: before your own scripts run, the scanner keeps a small diary of what the page listens for (which events it registers, and whether it tries to lock the screen orientation), because criteria such as motion actuation cannot be judged any other way. The diary records; it changes nothing, adds no behaviour and is not sent anywhere — it is read back by the scanner at the end of the visit, like the rest of the results. In the paid report the scanner also presses a few single letter keys with nothing focused, to see whether the page reacts to keyboard shortcuts the way a visitor would trigger them.

What is stored

A screenshot of the page, the HTML of the elements that fail a check, and technical details of the page load. Free-scan results are deleted after 30 days, research scans after 24 months; screenshots are never published. The privacy notice has the details, including what happens if a person is visible on a scanned page.