ComplyClear Accessibility

1.1.0

• Every place that reports how much of your site a scan covered now reports the same four numbers for that scan: how many pages it set out to check, how many it checked as visitors see them, how many it could only read from stored content, and how many it could not check. The scan history, the report, the emails and the documents you hand to someone else are counted once, by the scan itself, instead of each screen working it out for itself.
• Skip links are added only where they lead somewhere. ComplyClear now reads the finished page and adds a link for the main content, navigation, search or footer when your theme marks that area so a link can reach it, and adds none where it cannot. Each link appears when it receives keyboard focus and moves focus to the area it names. Before, four links were added to fixed names most themes do not use, and a script then hid them once the page had loaded.
• The links ComplyClear adds to your pages are no longer reported to you as problems with your site. The site-wide scan used to read them in the delivered page and report each one as a link pointing nowhere, on every page, and the plugin’s own styling could raise a focus warning as well.
• The setting for skip links now says what the feature does, including that a link is added only where the area it points to exists.
• A scan that could not read any page now says so instead of reporting a clean result, and the report says how many pages were checked, how many were reused from earlier checks, and how old the oldest reused check is.
• Colors in the Elementor panel were retuned so text, focus outlines and the controls themselves meet contrast in both the light and the dark editor themes, and the “Likely problems” badge in the block editor is legible against its background.
• Three checks now run the same way in the editor panels and the site-wide report, judged from one shared ARIA vocabulary. Broken id references: an aria-labelledby, aria-describedby, aria-controls, aria-owns, aria-activedescendant, aria-details, aria-errormessage, headers or label “for” attribute that points at an id the page does not contain, with every missing id listed, and an in-page link whose target is not on the page. Duplicate ids that something on the page references: a label, an ARIA relationship, a table header or an in-page link. Invalid ARIA: a role the WAI-ARIA standard does not define or that is abstract, an aria attribute it does not define, a value outside the attribute’s type, a required state missing from an explicit role, a container role without any of the child roles it requires, or a child role outside the parent it requires. Invalid ARIA is a new check on both engines. The two id checks reached the site-wide report in 1.0.9 and now reach the editor panels too, step for step. ComplyClear Accessibility Pro 1.1.0 adds a one-click fix for each of the three.
• In the block editor, an in-page link is now checked against the ids authored inside Custom HTML blocks as well as the blocks the panel inspects, and a heading’s HTML anchor counts as an id even though the editor canvas does not show it. When a page also holds content the panel cannot read at all, such as a shortcode or a synced pattern, a link whose target it cannot find is listed as a manual check that says so, rather than reported as a broken link.
• The site-wide scan and the document scan now run on sites that use plain permalinks. Before, the request they made carried a second question mark on such sites, WordPress answered that nothing was there, and the report said the scan did not finish.
• The block editor panel now reads a button’s link from the block itself. A button that points at an in-page target that does not exist is reported in the editor, where it can be fixed, and no longer only in the site-wide report.
• Every finding now states its own terms: which rule produced it and that rule’s version, whether it rests on a WCAG success criterion or is an advisory, whether it is a fact of the markup or a pattern for you to confirm, exactly what the rule does and does not test, which scan produced it and whether that scan finished, and the specific facts it was found on. The editor panel’s results name their rules too, and controls labeled by several elements at once are now named from all of them.
• Link and button names are now computed the same way on the server scan, in the editor panels and on the public scanner, and the order follows the accessible-name standard: a label referenced by aria-labelledby first, then aria-label, then the control’s own text, then a contained image or icon that carries a name, and the title attribute last. Two consequences you may notice: a link whose visible text is generic (“Download”, “Details”) is now reported even when a title attribute describes it, because the title is not what a screen reader announces as the name; and the editor panel now classifies the same finding the same way the report does.
• Every finding now carries a durable identity that survives edits to the page, so a decision you make about a finding in Pro (marking it reviewed, tracking it between scans) can follow the same finding instead of being lost when the page is saved or a builder regenerates its ids. The identity is computed from what the finding is about and where it sits in the page structure, never from the wording of the message. This release carries both the new identity and the previous one side by side, so nothing already recorded is dropped.

1.0.9

• New setting: delete the data this plugin stores when you delete the plugin. Off unless you turn it on, and it replaces the old instruction to ask your host to remove options by hand. Each plugin removes only its own data, so deleting Pro is what removes Pro’s history. Not supported on multisite, where nothing is deleted.
• New check: an element that points at an id which is not on the page. A label, description or relationship that references a missing id is ignored by assistive technology, and nothing was reporting it.
• New check: in-page links whose target is not on the page, reported as needs-review because many are created by scripts. A skip link that goes nowhere is the case worth checking first.
• Duplicate id findings are now limited to ids that something on the page actually references. A duplicate nothing points at cannot break a label, a relationship or a link, and reporting every one of them buried the few that mattered.
• New FAQ explaining why a scan result can lag behind a save, and how to set up a real scheduler if your host does not already provide one.
• The Elementor audit tab no longer stays on screen when Elementor hands the editor over to another mode. It was anchored to the panel’s edge and the panel had moved off screen, so the tab sat against an edge that was no longer there.

1.0.8

• The way back out of block settings is now always on screen. Clicking “Go to block and open settings” in the editor panel opens that block’s settings and offers “Back to Accessibility Audit” to return. On a shorter screen, or when the editor was already showing a notice, that control could be pushed below the visible area, so the way back existed but could not be clicked. It is now pinned to the bottom of the window, where neither the notice stack nor the window height can move it.
• Spelling is American throughout, including in this file.

1.0.7

• The site-wide scan now finishes on its own. Before, a scan of more than fifteen pages stopped part way and only continued if you clicked again in the same browser session, so a refresh sent it back to the start. It now runs through to the end, and a scan interrupted by a page reload picks up where it left off.
• The scan now covers your custom content types, not just posts and pages. Sites using types like events, meetings, agendas or staff profiles were being told coverage was complete while that content had never been looked at. Reports and exports now name which content types were checked and which were not.
• Findings are described by what you need to do about them. Where the plugin used to say “errors and warnings”, it now sorts everything into Confirmed problems, Likely problems, Manual checks and Suggestions, with a one-line explanation of each. The old wording described how sure the checker was, which is not the same question as what lands on your desk.
• The report opens with a ring for each of those four, so a page with nothing confirmed but twenty manual checks no longer looks finished at a glance.
• New installs now default to WCAG 2.2 AA, described as recommended practice rather than a legal requirement. WCAG 2.1 AA is still available and is the standard the DOJ Title II rule sets. Existing sites keep whichever standard they were already using.
• The editor panels no longer show a green all clear over content they could not read. Embeds, shortcodes, raw HTML blocks and similar are named as not checked, because a clean result over content nobody looked at is not a clean result.
• An image with no alt text is no longer reported as having a file name for alt text. The panel now reads the block’s own setting instead of whatever the editor happened to render.
• The report now says when a scan cannot support formal evidence. If your server cannot fetch its own pages, the scan still gives useful guidance, but it will say plainly that it cannot back an Accessibility Conformance Report or an export, instead of leaving you to discover that later.
• Pages whose fetch failed for a temporary reason are retried automatically in the background, so a busy server no longer costs you formal coverage.
• The accessibility statement no longer states things about your organization that the plugin cannot know. Staff training, procurement practice, testing you have performed, response times and the conformance claim itself are now marked as examples for you to keep, edit or delete, with an explanation at the top of the page.
• Corrected how Section 508 and ADA Title II are described. They are different laws covering different bodies, and the statement was presenting them as one requirement covering everyone.
• ComplyClear now has its own menu item instead of living under Settings. The Settings link still works.
• Jumping to a block from the block editor panel now opens that block’s settings, so the field you need is already in front of you.
• Page titles with an ampersand or curly quotes now display correctly in the report instead of showing the underlying character codes.
• Counts now read correctly when there is only one of something, and the report and exports now agree on what “a page with issues” means.
• Fixed a fault that made the block editor go blank as soon as the ComplyClear panel was opened. It affected every page, not just the panel.
• Fixed the ComplyClear admin pages loading without their styling or controls after the menu moved. The report would not display results and the scan buttons did nothing.
• New check: a button with no name a screen reader can announce. Links were already checked for this; buttons were not.
• New check: a page with headings but no H1. This is flagged as a suggestion rather than a failure, because no WCAG criterion requires an H1, but other checkers report it and now so does this one.
• A page using reCAPTCHA or hCaptcha is now flagged for review. It was being missed entirely, while gentler checks like Turnstile were reported.
• Advice about a skipped heading now names the levels actually found on your page, on the public scanner as well as in the plugin.
• Advice about an unlabeled form field now tells you where to set the label in the builder you are using, instead of describing HTML you cannot edit.
• Exports now use the same four categories as the rest of the plugin. The issue table in the printed report and the CSV were still saying Error, Warning and Notice.
• Every report, export and conformance draft now states that color contrast was not measured, because those are built from the HTML your server returns. The editor panels, which read the rendered page, say that it was.
• Summaries no longer roll Likely problems and Manual checks into one count, and now read correctly when there is exactly one of something.
• The conformance draft now describes one body of evidence per row. A criterion with five confirmed failures now says five pages, counting only the pages those failures were found on.
• Color contrast is checked again. A rule that was supposed to replace the old contrast check had been switched off by mistake since 1.0.6, so pages with hard-to-read inline colors were reported as having nothing to look at. It comes back as a manual check, because a scan of your server’s HTML cannot see colors set in a stylesheet and should not pretend otherwise.
• Small tap targets on icon-only links are flagged for review again, on sites set to WCAG 2.2. This was switched off by the same mistake.
• The Accessibility Report now shows your last completed scan when you come back to the page, instead of appearing empty until you run a new one.
• A measured color contrast failure is now reported as a confirmed problem. The editor reads the real colors from the page, so when the numbers fail there is nothing left to confirm. Cases the rules exempt, like disabled buttons and logos, stay listed as likely problems with a note saying why.
• Advice for an unlabeled form field now names the Label setting in the editor you are using, in the finding itself as well as the fix.
• Advice for tables now talks about tables. Two table findings were showing general advice meant for other kinds of content.
• Findings keep their identity between scans, so a card you expanded no longer swaps its advice for another finding’s after the page rescans itself.
• The Documents tab now uses the same four categories as the rest of the plugin, and counts of one read correctly.
• The Accessibility Report now shows whichever scan ran most recently, whether you started it yourself or it ran on a schedule. Before, a scheduled scan could leave the report describing an older manual scan while showing the newer results.
• If pages have been re-checked since the scan you are looking at finished, the report says so rather than quietly mixing them in.
• The report is tied to your last full scan. The printed and CSV exports and the conformance draft describe every published page as it stands right now, and each of them says so, because after re-checking a single page those are different questions with different answers. Every published page stays counted in the exports and the draft even when its last check failed, so a coverage gap is always visible rather than quietly dropped.
• The editor panels now show all four categories at once, including any that are empty. Manual checks in particular used to disappear when a page happened to have none, which made it look like an occasional extra rather than something always waiting for a person.
• Each finding now carries its category, criterion number, level and criterion name on one line, and the block editor panel names the criterion and level it was previously unable to show at all.
• The Elementor panel button has moved off the middle of the page, where it sat over the design you were editing. Its counters now name what they count, and every category a scan finds is visible on the button instead of only two of them.
• The Elementor panel now narrows on a small screen instead of covering the page it is describing.
• The explanations of what was and was not checked in the editor panels now sit behind a heading you can open, rather than taking up permanent room.
• The editor panel explains why one measured contrast failure is a confirmed problem and another is a likely problem. The difference is a rule exception that cannot be settled automatically, and the panel now says so instead of appearing inconsistent.
• The Auto-scan checkbox now matches the panel it sits in. It was drawn by the operating system, so on a dark panel it could appear as a bright square that read as a color swatch rather than a checkbox.
• Opening a block’s settings from the block editor panel now offers a way back to the audit, and your results are kept while you are away.
• Saving alt text from the Elementor panel now tells you first that it saves to the Media Library, applies wherever that image appears, and is not part of the page’s own save or undo. It checks whether the change could be undone and tells you before you save, instead of promising an Undo and only afterwards admitting it was not possible. When it is possible, the Undo stays in the panel until you use it or dismiss it, so live auto-scan refreshing the results no longer takes it away before you can reach it.
• Adding a table caption from the block editor panel now clears the caption finding in the same session. The caption was being saved correctly all along, but the panel could not see it in the editor canvas and went on reporting the same problem, which left no way to tell a working fix from a broken one.
• Spelling is now consistent American English throughout, matching the rest of the plugin’s wording.

1.0.6

• Accuracy pass. Several checks were reporting problems that were not actually there, and a few were reported with more certainty than a page scan can support. Findings are now labeled by how well they can be established, so a real failure is never confused with something that just needs a look.
• Every finding now says whether it is a confirmed failure, something that needs review, a manual check, or advisory guidance. Confirmed means the page itself shows the problem. Nothing else is presented as a failure.
• Removed checks that were applying the wrong standard: a CAPTCHA on an ordinary page is not an Accessible Authentication failure, a page without a help link is not a Consistent Help failure, and a form with several fields is not evidence of Redundant Entry.
• A missing table caption and an unwarned new-window link are now advisory notes rather than failures. Both are recommended techniques, not requirements.
• Color contrast is now a manual check. The previous check guessed at the background when it could not see one, which produced false failures on dark sections and quietly skipped light text entirely. It now tells you what to check instead of guessing.
• Tap target size is now a manual check. It was estimating from link text length rather than the rendered size, which is not something a page scan can measure.
• Decorative images marked with role=”presentation” or aria-hidden are no longer reported as missing alt text. An image with empty alt that is the only content of a link is still reported, because that link has no name.
• Hidden form fields, including spam honeypots, are no longer reported as unlabeled. They are not shown to anyone.
• Timed refresh: a redirect with no delay is no longer flagged, and very long delays are exempt as the standard intends.
• Fixed a missed failure: a link whose aria-labelledby pointed at an element that does not exist was treated as if it had a name. That is a real problem for screen reader users and was being skipped.
• The numeric score has been removed for now, since it was calculated from the checks corrected above. Findings and coverage are shown instead.

1.0.5

• Fixed the “Report an Issue” contact form in the accessibility widget, which could not submit. Reports now send properly, show a confirmation message in place, and work even with JavaScript turned off. If your site shows this widget with the contact form, please update.
• Vague link text such as “Read more” is no longer flagged when the link has a descriptive accessible name from aria-label, aria-labelledby, or a title attribute. This matches how screen readers announce the link and how the in-editor check already judged it, so the site report and the editor now agree.
• Hardened the PDF document scanner against malformed files: a damaged or crafted PDF can no longer crash the media uploader, and files that cannot be read are now reported honestly as “could not be checked” instead of failing silently.
• PDFs saved with incremental updates are no longer misread as scanned, image-only documents.
• Code snippets in the site-wide report now display correctly instead of showing raw HTML entity codes.
• Word (.docx) documents are now scanned with the same safeguards as PDFs, and unusually large or highly compressed files are handled gracefully instead of straining the server.
• The plugin no longer adds its own viewport tag, which could duplicate the one your theme already sets. Skip links now hide automatically when their target is not present in the theme, so a visitor never lands on a broken skip link.
• The Documents report and the Media Library accessibility column are now more accurate. A file that changed after it was scanned, or whose type no longer matches, is shown as needing a rescan instead of keeping an old result.
• Very large or unusually complex pages now report as incomplete with guidance, rather than returning a partial score, so a page that could not be fully checked is never mistaken for a clean pass.
• Additional security hardening in document parsing and the Elementor editor preview.