Loading legal page…
Last updated: 18 August 2026
This statement covers the CrowAgent application at app.crowagent.ai. Our marketing website at crowagent.ai has its own separate statement. Neither one implies anything about the other, because they were tested separately and the results are different.
Partially conformant with WCAG 2.2 level AA.
The word partially is doing real work. Automated testing across the application finds no critical, moderate or minor issues, and we have now also completed two of the manual checks that automated tools cannot perform. But automated tools detect only a portion of the WCAG success criteria, we have not tested with screen readers, and we have not tested with disabled users. Claiming full conformance on the evidence we hold would overstate what we have actually measured.
Testing was carried out on 18 August 2026. The application is behind a login, so the signed-in parts of it were measured on our staging environment rather than on production. Staging runs the same application code as production; it does not hold the same data.
| Measure | Count | Notes |
|---|---|---|
| Page routes in the application | 142 | Counted from the source tree, not from a hand-written list. |
| Routes visited by the automated sweep | 135 | Every route except the seven described in section 4. |
| Routes successfully measured | 106 | Scanned in both the dark and light themes, so each measured route was scanned twice. |
| Routes visited but not measurable | 29 | Described in section 4. They are counted, not discarded. |
| What we tested | How | Result |
|---|---|---|
| Automated checks, every severity | axe-core driven through a real browser, signed in, both themes | 0 critical, 0 moderate, 0 minor. Serious findings were confined to colour contrast in the light theme, on pages that serve only the dark theme, so no user can reach them. |
| Keyboard-only operation | Real Tab key presses through every focusable control on each measured route | No keyboard traps. Every control the page exposes was reachable. Four fixes were made to the “skip to main content” link, described below. |
| Text resizing and reflow at 400% | Measured at a 320 CSS pixel wide viewport, the equivalent of zooming a 1280 pixel window to 400% | No route required horizontal scrolling. Wide tables scroll inside their own labelled region, which WCAG permits. |
The instruments were themselves checked before any result was believed. The scanner was given a deliberate fault on a live page and had to name it; the reflow measurement was given a deliberately over-wide element and had to catch it; and the sweep proved it was genuinely signed in before it measured anything, by confirming that a signed-out visitor is turned away from the same page.
The keyboard testing found that the “Skip to main content” link, which is the first thing a keyboard user reaches on every page, did not work on thirteen routes. On eight of them it pointed at a target that did not exist; on five the target existed but focus did not move to it, so pressing the link left the user exactly where they were. Both faults have been corrected.
We list these rather than quietly reducing the denominator, because the size of what was not tested is part of an honest result.
In total 29 routes were visited but could not be measured, and 7 more could not be visited at all. Those 36 routes are counted in the table above rather than removed from it, because a coverage figure that quietly drops what it could not test is not a coverage figure.
Their absence is the reason section 1 does not claim full conformance:
Until those are done there may be barriers in this application that our testing has not surfaced. If you hit one, we would rather hear about it than have you assume we already know.
If any part of this application prevents you from doing something, tell us at hello@crowagent.ai. Please include the page address and what you were trying to do. We aim to respond within 5 working days. If you need something here in a different format, ask and we will tell you honestly whether and when we can provide it.
The Public Sector Bodies (Websites and Mobile Applications) (No. 2) Accessibility Regulations 2018 apply to public sector bodies. CrowAgent Ltd is a private supplier and is not a public sector body, so those regulations do not bind this application. We publish this statement because accessibility matters to the people who use CrowAgent, and because our public sector customers reasonably ask suppliers for one, not because we are legally required to.
Our obligations under the Equality Act 2010 to make reasonable adjustments do apply, and we take them seriously. If you are dissatisfied with how we respond to an accessibility complaint, you can contact the Equality Advisory and Support Service (EASS).
This statement was prepared on 18 August 2026 and reflects testing carried out on that date, using our own tooling as described in section 2. It has not been independently audited by a third party, and we will say so here until it has been.
We will update this statement when the outstanding testing is complete, and we will change the conformance claim only when the evidence changes, not before.