Developer Agents
Identify common accessibility barriers on your site and get a prioritized list of fixes to review with your team.
Paste a URL and the agent scans for the issues that most often lock people out: low-contrast text, missing alt text, broken heading structure, unlabeled form fields and keyboard traps. Findings come back grouped by severity with remediation guidance, so your team knows what to review first.
Try an example
Roughly one in six people lives with a disability, and a large share of websites still present basic barriers: text too faint to read, images without descriptions, forms that cannot be completed with a keyboard. A website accessibility audit is how you find out which of those barriers exist on your site — before a user, a client or a complaint finds them for you.
This agent is for developers who want a fast first pass before a release, designers checking that a palette survives contact with contrast requirements, site owners who inherited a site and do not know its state, and agencies that need a concrete accessibility deliverable for clients.
You provide a URL. The agent examines the rendered page for the issues automated analysis can reliably detect: color contrast ratios, missing or unhelpful alternative text, heading hierarchy problems, form fields without labels, links with meaningless text, missing landmarks, focus indicators and elements that suggest keyboard traps. Each finding includes a reference to the relevant WCAG success criterion so your team can look up the exact requirement.
Honesty matters here, so two boundaries are explicit. First, the report helps identify issues for human review and remediation — it does not, and no automated tool can, certify that a site 'is compliant'. Conformance with WCAG and legal obligations involve judgment calls and assistive-technology testing that automation cannot perform. Second, the agent deliberately avoids the dark pattern of overlay widgets that promise instant compliance; the output is a work list for fixing your actual markup.
If the audited site is a Dual7 project, remediation is short: adjust contrast tokens, add alt text, repair heading structure and label forms in the same place the page lives, then republish. For organizations that need a documented review trail, fixes can move through the governed pipeline with sign-off at each stage.
Contrast, alt text, headings, labels and keyboard access are checked on every run — the categories behind most real-world accessibility failures.
Issues are grouped so barriers that block tasks entirely — an unreachable form, a keyboard trap — are reviewed before cosmetic ones.
Each finding cites the relevant success criterion, so remediation discussions start from the actual requirement, not opinion.
Findings explain what a fix looks like — a contrast ratio to hit, the alt text to write, the label to add — so the report doubles as a work list.
The report identifies issues for review and is explicit that automated scanning cannot replace manual and assistive-technology testing.
Run the audit before and after remediation to confirm fixes landed and nothing regressed in the same release.
Developers get element-level detail, designers get contrast and focus findings, and owners get a plain-language summary of where the site stands.
Point the agent at any public page — homepage, checkout, signup flow or a page you suspect has problems.
Contrast, alt text, heading structure, form labels, link text, landmarks and focus behavior are examined against common WCAG failure patterns.
Read the grouped report: what the issue is, which criterion it relates to, and what remediation would involve.
Fix the markup, publish, and run the audit again to confirm the findings are resolved and nothing new appeared.
Text and interactive elements are checked against contrast ratio requirements, with the failing combinations listed so designers can adjust tokens rather than guessing.
Images are flagged when alt text is missing, empty on a meaningful image, or filled with placeholder junk like filenames — with guidance on writing useful descriptions.
The outline is examined for skipped levels and headings used for styling instead of structure, which is how screen reader users navigate the page.
Inputs are checked for associated labels, error messages are checked for being perceivable, and required fields for being identified without color alone.
Interactive elements are checked for focus visibility and the page for patterns that suggest keyboard traps or unreachable controls.
Links with vague text like 'click here' are flagged, and the page is checked for the landmark regions assistive technology uses to jump between sections.
Findings are compiled into a single report grouped by impact, each with its WCAG reference and a concrete remediation suggestion for human review.
A list of detected issues with the element or pattern responsible, grouped by severity for triage.
The specific text and background combinations that fall below ratio requirements, ready to hand to a designer.
Images, inputs and controls missing accessible names, with guidance on what each should say.
Heading outline problems, missing landmarks and keyboard access concerns with references to the relevant criteria.
The findings reordered as a practical sequence — blocking barriers first — for a team to review and work through.
Example output
Input: a company's homepage and signup flow. Findings: light-gray body text failing contrast requirements, hero images with filename alt text, a signup form whose fields are labeled only by placeholder text that disappears on focus, and a mobile menu that cannot be opened by keyboard. Report: fourteen issues grouped by severity, each with its WCAG reference and a suggested fix for the team to review.
A front-end developer runs the audit on a staging URL before launch so contrast and labeling issues are caught while they are cheap to fix.
A designer audits key pages after a brand refresh to confirm the new colors still meet contrast requirements on real text.
A new owner who inherited a website runs the audit to learn its accessibility state before planning any other work.
An agency runs the audit as the automated first pass of a broader accessibility engagement, then verifies findings manually.
A vendor whose customers ask about accessibility in RFPs audits its site and builds a remediation plan it can speak to honestly.
A nonprofit audits its donation page so supporters using screen readers or keyboards can actually complete a gift.
A product team scans its signup and onboarding pages, where unlabeled fields and focus problems directly cost conversions.
A university web team audits high-traffic program pages as the first step of a campus-wide accessibility effort.
Stores find the checkout and product-page barriers that block purchases for customers using assistive technology.
Schools and universities, which often serve users with a wide range of needs, identify issues across large page inventories.
Clinics and providers check that appointment booking and patient information pages are perceivable and operable by everyone.
Public bodies with formal accessibility obligations use the audit as an automated first pass alongside manual review.
Banks and fintechs audit application and account flows where a barrier can lock a customer out of their own money.
Product teams check marketing and signup pages where accessibility problems quietly suppress conversion.
Organizations whose mission includes inclusion make sure their own donation and program pages live up to it.
You provide
https://our-marketing-site.com
The agent
The agent scans the homepage for contrast, alt text, heading structure, landmarks and keyboard access.
You get
A report with eleven findings: three contrast failures, four images needing alt text, a skipped heading level and a footer with no landmark — each with a WCAG reference.
You provide
https://checkout.example.com — audit our checkout flow
The agent
The agent examines the checkout page's form fields, error handling and focus behavior.
You get
Findings showing fields labeled only by placeholders and errors announced by color alone, with remediation guidance for the team to review.
You provide
We just rebranded — check that our new palette still works
The agent
The agent checks text and button contrast across the homepage and key templates using the new brand colors.
You get
A contrast failure list naming the exact color pairs that fall below requirements, so the designer adjusts two tokens instead of the whole palette.
You provide
Audit our donation page before our annual campaign
The agent
The agent scans the donation form for labels, keyboard operability and perceivable error states.
You get
Confirmation that the form is largely sound, plus two flagged issues — a vague 'submit' link and a missing error summary — fixed before launch.
You provide
New client site, we inherited it — what state is it in?
The agent
The agent audits the homepage, contact page and top landing pages to establish a baseline.
You get
A severity-ranked baseline report the agency uses to scope remediation and re-audit against after each round of fixes.
You provide
Scan our docs — developers say the code blocks are hard to read
The agent
The agent checks contrast and zoom behavior on documentation pages, including syntax-highlighted code.
You get
Findings confirming the code theme fails contrast in three token colors, with accessible alternatives suggested for review.
| Aspect | With the agent | Manual process |
|---|---|---|
| Getting started | Paste a URL and get findings in one run | Set up testing tools and learn each one's output format |
| Contrast checking | All failing combinations listed together | Sampling colors one pair at a time with an eyedropper |
| Alt text review | Missing and placeholder alt text flagged across the page | Inspecting every image individually |
| WCAG mapping | Each finding cites the relevant success criterion | Cross-referencing issues against the guidelines yourself |
| Prioritization | Findings pre-grouped by severity | You triage a raw issue list from scratch |
| Re-testing after fixes | Re-run the same audit and compare | Repeat the whole manual pass and hope nothing was missed |
| Honest scope | States plainly what automation cannot verify | Overlay widgets that claim instant compliance without fixing markup |
A chatbot can critique markup you paste in, but it cannot measure the contrast of your actual stylesheet or test your live heading outline. The agent reads the real page.
A consistent rule set means results are comparable over time — you can tell a fix worked because the same audit no longer reports it.
Instead of a conversational opinion, each issue carries a success-criterion reference your team can verify against the actual guidelines.
The report is structured for triage — severity, element, fix — so it can be assigned and tracked rather than scrolled through.
The agent identifies issues for review and says so. It does not overlay a widget and declare the site compliant, which protects you from false confidence.
Automated scanning reliably catches a subset of issues. Follow it with keyboard testing and, where possible, testing with screen readers before drawing conclusions.
A form that cannot be completed by keyboard matters more than a skipped heading level. Work the severity ranking in order.
If body text fails contrast, change the token once rather than patching individual pages — the fix propagates everywhere the token is used.
Describe what the image communicates in context, not what it literally contains. A chart's alt text is the takeaway, not 'a chart'.
Placeholders disappear when someone types and are inconsistently announced. Every field needs a persistent, associated label.
Tab through your key flows yourself. Ten minutes of firsthand experience teaches more than any report about where focus gets lost.
New themes and components reintroduce old failures. Run the audit after each significant visual change, not once a year.
A documented history of findings and remediation is genuinely useful — for procurement questions, for accessibility statements, and for your own regression checks.
This agent examines the issues automated analysis can reliably detect: color contrast, missing or poor alt text, heading hierarchy problems, unlabeled form fields, vague link text, missing landmarks and patterns that suggest keyboard traps. Each finding references the relevant WCAG success criterion.
No — and no automated tool can. The audit helps identify common issues for human review and remediation. Legal conformance involves judgment, manual testing and assistive-technology evaluation that automation cannot perform. Be wary of any tool that claims otherwise.
No. Automation is strong at measurable problems like contrast and missing attributes, but many criteria — whether alt text is meaningful, whether focus order is logical, whether content is understandable — need human judgment. Use the report as a thorough first pass, then verify manually.
Findings are mapped to WCAG 2.2 success criteria — the current W3C recommendation — so your team can read the exact requirement behind each issue. The report names the criterion per finding rather than asserting a blanket conformance level for the whole site.
Developers before releases, designers validating color and focus choices, site owners assessing an inherited property, and agencies producing an accessibility deliverable — anyone who needs a concrete, prioritized picture of a site's common barriers.
Work it as a triage list: fix blocking barriers first, apply design-token level fixes where a failure is systemic, and re-run the audit to confirm. If the site is a Dual7 project, many fixes can be made and republished in the same place.
Overlays inject a toolbar and claim the underlying site is now accessible, which the accessibility community has repeatedly shown is not the case. This audit identifies issues in your actual markup so you can fix the site itself — the only approach that genuinely helps users.
The agent reads public URLs. For gated flows, audit a staging environment or the public entry points, and test the authenticated journey manually with a keyboard and screen reader.
After every redesign or major component change, and periodically in between — content editors routinely introduce new issues like missing alt text. Treating accessibility as a recurring check rather than a one-time project is what keeps a site accessible.
Several overlap — descriptive alt text, proper headings and clear link text are useful to search engines as well as assistive technology. Accessibility is worth doing for its own sake, but the work rarely hurts discoverability.
Dual7 is in early access, which is free with no credit card required. Run the audit on your URL and review the full findings before deciding anything.
Dual7 Agents
Identify common accessibility barriers on your site and get a prioritized list of fixes to review with your team. Start with your own input — the output is an editable project you own.
500+ users in production · Kanan.co moved off Salesforce onto Dual7 in four months · 100% code owned
Dual7 · a product of AlgorithmShift · Build once. Own forever.