Accessibility

We want everyone to be able to read this site and use everything on it. This page says what we did, how we checked it, and what we know is still imperfect. If something does not work for you, tell us and we will fix it.

The standard we build to

This site is built to meet WCAG 2.1 Level AA, the international accessibility standard. We also test against the newer WCAG 2.2 Level AA rules.

The Americans with Disabilities Act does not list technical rules for websites. When the U.S. Department of Justice explains what an accessible site looks like, it points to WCAG. So WCAG 2.1 AA is what we measure ourselves against, and WCAG is what this page reports on.

What we did

  • Everything works with a keyboard. You can reach every link and button with Tab, and a “Skip to content” link lets you jump past the header and menu.
  • The menu is plain links. It sits in a labeled navigation landmark, with no drop-downs and nothing that opens or closes. The page or section you are in is marked for screen readers, and on screen it is underlined as well as darker, so the cue does not depend on color. On a narrow screen or with large text, the menu wraps onto a second line rather than hiding behind a button.
  • Breadcrumbs show where a page sits. Pages deeper in the site start with a short trail of links back up, in its own labeled navigation landmark.
  • The focus outline is always visible. Whatever you are on gets a clear teal outline, measured at 6.2:1 against the page in light mode and 9.5:1 in dark mode.
  • Text contrast is measured, not guessed. Body text is 17:1 against the page background. The lowest-contrast text anywhere is 5.4:1. The standard asks for 4.5:1.
  • The page is built from real structure. Headings are in order, lists are real lists, table headers are marked as headers, and the main content sits in a main landmark. Screen readers use this to let you jump around instead of reading top to bottom.
  • Text can grow and the page still works. Zoom to 200%, or use a 320-pixel-wide screen, and nothing is cut off and nothing scrolls sideways.
  • Nothing moves on its own. No animation, no autoplay, no carousels, no pop-ups, no timeouts.
  • Dark mode is tested too, not just light mode. The page opens in whatever your device asks for, and the sun-and-moon button in the header switches it if you want the other one.
  • The theme button is icon-only, so it carries a text label for anyone who cannot see the icon. Screen readers announce “Switch to dark mode”, and that label changes with the button. It is reachable by Tab and works with Enter or Space. The icon measures 5.4:1 against the page, above the 3:1 the standard asks for, and the button is 32 by 32 pixels, above the 24-pixel minimum target size.
  • The tooltip on that button can be dismissed with Escape and says the same words the screen reader announces, so nobody gets a different message depending on how they are reading the page.
  • Images are described. The logo has a text alternative; decorative copies of it are hidden from screen readers so you do not hear it twice.
  • The site works without JavaScript. Script does two small jobs: it assembles our email address to keep it away from spam robots, and it runs the theme button. With script off, the address is still written out for you to read, and the theme button hides itself rather than sitting there doing nothing — dark mode still follows your device.

How we checked

  • Automated testing. Every page was tested with axe-core 4.10.2 against the WCAG 2.0 A/AA, 2.1 A/AA and 2.2 AA rule sets, plus Deque’s own best-practice rules. All 27 pages return zero violations, in light mode and in dark mode.
  • Contrast measured by calculation. Every text and background pair was computed to the WCAG formula rather than eyeballed, and so were the borders of buttons, which automated tools do not check.
  • Reflow and zoom. Each page was rendered at 320 pixels wide and checked for anything overflowing or clipped.
  • Text spacing. We applied the spacing the standard requires us to survive — taller lines, wider letters and words, more space between paragraphs — and checked that no text was lost or overlapped.
  • The theme button was tested by what it actually draws, not by what the code sets. An early version looked correct in testing while showing every icon at once on screen, because the test asked the code rather than the page. Checks now read the rendered result.

What the testing found, and what we changed

Being transparent about this matters more than claiming it was right the first time. It was not. When the site was first built, automated testing found two problems and our own manual checks found three more. All five are fixed.

  • The main content was not inside a main landmark, so screen reader users had no quick way to skip to it. Added, along with a skip link.
  • The outlined “Read the documents” button was drawn with a border measuring only 1.68:1 against the page. The standard requires 3:1 for the edge of a control. It is now 5.44:1.
  • Wide data tables were clipped on narrow screens. They are now scrollable and, importantly, reachable by keyboard rather than mouse only.
  • Some lists were styled in a way that makes Safari and VoiceOver stop announcing them as lists. Fixed so the list structure survives the styling.
  • Table header cells were not explicitly marked as column headers.

Testing the redesigned site in October 2026 found three more, also fixed:

  • With script turned off, the theme button showed but did nothing. This page said it hid itself; a style rule was overriding that. It now hides.
  • On the fundraising events guidance, each section’s whole text was marked up as one long heading, which made the page very hard to navigate by headings, and its table of contents links led nowhere. The sections are now headings followed by ordinary text and lists.
  • On a 320-pixel screen with the larger text spacing applied, the home page heading and some long package names in code pushed the page sideways. They now wrap.

What is still imperfect

We would rather say this plainly than let you find out yourself.

  • Wide tables scroll sideways on a phone. A table of retention periods needs rows and columns to make sense, so we made it scrollable instead of breaking it apart. It is keyboard reachable, but a single-column version would be easier.
  • Code examples scroll sideways on a phone. A line of code cannot be rewrapped without changing what it means, so a long line scrolls inside its own box. Each box can be reached with Tab and scrolled with the arrow keys.
  • No independent audit yet. All testing so far is our own. We have not paid for a third-party review.
  • No testing with disabled users yet. We have tested with tools and by keyboard. That is not the same as watching someone use a screen reader in earnest, and we know it.
  • Automated tools catch only part of the picture. Zero violations from axe-core means no machine-detectable failures. It does not mean the site is perfect.

Telling us about a problem

Write to info [at] automatethe.cloud and say what page you were on, what you were trying to do, and what happened. If you use a screen reader, a magnifier, voice control or anything else, telling us which one helps us reproduce the problem.

We will reply. If a fix is quick we will make it and tell you; if it is not, we will tell you what we are doing and roughly when.

If you would rather not email, anything in our other contact routes works.

About this statement

Written on October 2, 2026, and updated on October 22, 2026, when the site was redesigned, covering every page of this site. It describes testing we performed ourselves. It is not a legal certification, and no one has certified this site on our behalf.