About our scanners

Brian Beals, LLC runs two automated readers. Each one names itself in its User-Agent and links to this page. Here is what each does, how fast it goes, and how to make it stop.

BrianBealsLLC-A11y

The accessibility scanner. It loads a page in an ordinary browser engine and measures it against WCAG 2.1 A and AA: contrast, names on links and controls, headings, how the page reflows at phone width, and the PDFs the page links. The same name is on four kinds of request.

  • Recon: a short look at a handful of public pages, ten at most and usually four to six, plus up to six of the PDFs they link. It's started by hand, one site at a time.
  • Evaluation: the pages a site owner has agreed to in a signed engagement, during a window agreed with their team.
  • Document probe: reads PDFs for tags, language, title, and form fields, either the documents linked from pages recon looked at or, for a document inventory, every document on a site. It runs either as part of recon or under a signed engagement, and says which each time.
  • Site check: while a list of organizations is being built, one or two requests to an organization's home page to confirm a web address belongs to it.

How fast

  • One browser tab, one page at a time, for recon and evaluation.
  • Recon waits at least one and a half seconds between pages, longer if your robots.txt sets a longer Crawl-delay.
  • An evaluation loads no more than one page every three seconds and fetches documents one at a time, at least a second apart.
  • The document probe reads one PDF at a time, at least one and a half seconds apart, longer under recon if your robots.txt sets a longer Crawl-delay.
  • A site check asks each home page for itself once or twice; different organizations' sites can be checked at the same time.
  • A page that scrolls sideways is loaded a second time at phone width.

robots.txt

Recon reads robots.txt before every page and document and skips anything disallowed, whether the rule names BrianBealsLLC-A11y or applies to all agents. An evaluation covers only the pages agreed in scope with the site owner, so anything to leave out goes into that scope. The document probe follows robots.txt the same way as recon when it runs as part of recon; under a signed engagement with an agreed document list it fetches that list as given. A site check asks only for the home page and doesn't read robots.txt.

User-Agent

Desktop, and every document request

Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/127.0.0.0 Safari/537.36 BrianBealsLLC-A11y/1.0 (+https://brianbeals.com/scanner)

Phone-width check

Mozilla/5.0 (iPhone; CPU iPhone OS 17_5 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/17.5 Mobile/15E148 Safari/604.1 BrianBealsLLC-A11y/1.0 (+https://brianbeals.com/scanner)

BrianBealsLLC-NoticeCheck

A reader looking for one page: a notice federal rules require an organization to publish on its website. For a health care organization that's the Section 1557 notice of nondiscrimination. For a public entity it's the ADA coordinator and grievance procedure. It reads the sitemap, tries the few addresses such a notice usually lives at, and reads the home page for a link to one. It runs no accessibility tests and records no findings about the site.

How fast

  • Started by hand, one site at a time.
  • Four seconds between requests.
  • At most 40 requests to a site, or 75 for the public-entity check.

robots.txt

It reads robots.txt first and skips any address disallowed, whether the rule names BrianBealsLLC-NoticeCheck or applies to all agents. Its four-second pause is fixed and doesn't change with Crawl-delay.

User-Agent

Every request

Mozilla/5.0 (compatible; BrianBealsLLC-NoticeCheck/1.0; +https://brianbeals.com/scanner)

What neither one does

  • Anything other than read-only GET requests.
  • Submit a form.
  • Try to log in, or try to get past a CAPTCHA or a challenge.
  • Load testing or vulnerability scanning.
  • Retry a refused page. A refusal is recorded and the reader moves on. The one exception is the document probe: a PDF refused to a plain request is asked for once more through the browser engine, then recorded either way.

Stop or block

Email or call and I'll stop scanning your site. Write to brian@brianbeals.com or call (941) 979-6282.

To block either one yourself, refuse requests whose User-Agent contains BrianBealsLLC at your firewall or CDN, or disallow them by name in robots.txt:

User-agent: BrianBealsLLC-A11y Disallow: / User-agent: BrianBealsLLC-NoticeCheck Disallow: /

robots.txt stops recon and NoticeCheck. To change what a contracted evaluation covers, tell me and the scope changes. A client who wants the evaluation let through instead will find what to send their IT team on before we start.