Statement of conformance
Accessibility
Two pieces of software are covered here, and they are not the same thing: the Capstan application, which your team would use every day, and this site, which you are reading now. They are built by the same people to the same standard, so this statement keeps them apart and says plainly which claim covers which.
This site, by what you use to read it.
Everything in this section is checkable against the page you are on, which is the only kind of accessibility claim worth publishing. Where a row has a limit, the limit is in the row.
- If you navigate by keyboard
-
The skip link is the first thing Tab reaches on every page, and one press takes focus into the main content. One focus ring is defined once for the whole site, and no page here is allowed to draw a second one or turn it off. The header menus are native disclosure elements, so they open and close from the keyboard with no script at all, and a small script closes them on Escape.
Nothing on this site traps focus, because nothing on this site is a dialog.
- If you use a screen reader
-
Every page carries the same landmarks, a header, a navigation, a main and a footer, with one h1 and real headings under it. Tables are tables, breadcrumbs are a labelled navigation, and every image carries a text alternative, including the placeholders that say in words that the product screenshot is still pending. Each page is an ordinary page load rather than a routed view, so you start at the top of a new document instead of somewhere in the middle of the old one.
This is the claim we can least prove. Nobody has worked through this site with a screen reader, so if something here reads wrongly, we will not know until you tell us.
- If you enlarge the text, or work in a narrow window
-
Every built page is measured at twelve widths, from 320px to 1440px, before a change ships. At none of them may the page scroll sideways, may a box run past the edge, or may text be cut off by a fixed height. 320px is the width WCAG 2.1 asks about under 1.4.10 Reflow, which is why the sweep starts there. Anything genuinely too wide for a phone, a long comparison table, scrolls inside its own box instead of moving the page.
- If you point with a finger, or with limited accuracy
-
Standalone controls are at least 44px tall at touch widths. AA asks for 24 and AAA asks for 44, and the check fails under 24 and warns between 24 and 44, so a control that merely passes still gets looked at. Two things are deliberately left alone: a link inside a sentence, which the rule itself exempts and which would wreck the line spacing if it were padded, and a checkbox, which stays its own size while its label carries the tap area.
- If you need contrast
-
The text colours are computed against the grounds they sit on rather than judged by eye: primary text on the page ground, 14.02:1; secondary text on the page ground, 5.70:1; primary text on a tinted band, 12.86:1; secondary text on a tinted band, 5.23:1. All of them clear the 4.5:1 that AA asks for on body text. Nothing here is told apart by colour alone: no heading, card or band on this site is coloured by state, and where a message reports a state, the message says so in words.
This site ships one theme, and it is the light one. There is no dark mode here to verify.
- If you ask your system for less motion
-
Nothing on this site moves on its own. There is no video, no carousel and no autoplaying anything, and the two effects that do exist, a scroll-driven fade on some figures and the skip link sliding into view, both switch off when your system asks for reduced motion, along with smooth scrolling.
- If you browse with JavaScript off
-
Every content page renders in full, because it is static HTML. The menus and the FAQ blocks are native disclosure elements, and the one script the chrome loads does nothing but close a header menu, on Escape, on a click outside it, or when you open another one. The exceptions are the eight free tools and the estimator on the pricing page, which compute in your browser and need scripting to produce a result. They send nothing anywhere and store nothing.
The product, which is a different piece of software.
The Capstan application is built to the same WCAG 2.1 AA baseline, and that baseline is guarded by a test rather than by good intentions.
- A skip-to-content link and a clear landmark structure on every screen.
- A visible focus outline on every interactive control.
- Focus moved to the main content on navigation, so keyboard and screen-reader users land on the new screen rather than at the top of the menu.
- Form fields with real labels, error messages tied to their field, and status messages announced politely.
- Dialogs that trap focus, close on Escape, and return focus where it started.
- Colour contrast checked against the palette, so text meets the AA ratio rather than only looking fine.
How that is checked, and how far the check reaches
An accessibility scanner runs inside the end-to-end suite, tagged for WCAG 2.0 and 2.1 at levels A and AA, and fails on any serious or critical finding, so a regression stops the work rather than waiting for somebody to report it. It covers four surfaces: the sign-in page, the dashboard shell, the Webhooks developer page and an admin configuration surface.
They are named rather than summarised because the honest reading of that sentence matters more than the impressive one: four surfaces are not the whole application, and an automated scan is not an audit. The next section says both in as many words.
What this page does not claim.
A statement that hides its limits is worth less than no statement at all, because the first limit a reader finds for themselves discredits everything above it.
-
No independent audit, of either one
Nobody outside this company has assessed the product or this site against WCAG. That is why the word on this page is baseline: neither is certified and neither is compliant, and we do not use either word about ourselves.
-
An automated scan proves a subset
A scanner reads the machine-checkable part of WCAG and stops there. It can see a control with no label. It cannot see a heading that misdescribes what follows it, a focus order that makes no sense to a person, or an alternative text that is technically present and useless.
-
No assistive-technology pass
Nobody has worked through the product end to end with a screen reader, a magnifier or voice control. Until somebody has, that is a gap in what we know rather than a clean result.
-
Contrast is verified for one theme
A dark theme is planned for the product, and only the light theme is verified today. This site ships a single theme, so there is nothing further to check here.
-
The target is WCAG 2.1 AA, not 2.2
Where we hold a 2.2 line anyway, the 44px control being the obvious one, we do it because that is the size a thumb actually hits, not so that we can claim the newer standard.
-
No VPAT, and no conformance report
We publish neither. A report summarising an automated scan would imply the audit this page has just said has not happened, and the implication is the whole value of the document.
When an independent audit is completed, this page will say so, with the date and with what it found. Until then, the date at the top is the date somebody last checked every sentence here against the code and the product documentation.
If something here blocks you.
A barrier is a defect, not a suggestion, and reports are how we find the failures an automated scan is structurally unable to see.
Write to support@usecapstan.com, which is the address for something that is not working. If another one fits better, the contact page lists every address we have. Three things make a report reproducible, and none of them is required:
- The page or screen, and what you were trying to do when it stopped you.
- Your browser, and the assistive technology you were using, if you know it.
- What you expected to happen instead.
We publish no support response time anywhere on this site, and this page will not invent one for itself. What we will do is treat the report as a defect, and where the fix is a rule rather than a patch, it goes into the checks that run before the next change ships.
Found something here you cannot use? That is the report we most need.