Use Cypress tests to visit meaningful application states, run automated accessibility checks against those rendered states, and add explicit assertions for behavior and content your application requires. AI agents can help author tests, inspect Cypress failures and accessibility trees, and verify proposed fixes—but a clean scan is not proof that an application is accessible.
Build an accessibility-testing workflow around real application states
A scanner can only inspect what your test renders. Start by identifying important pages and interactive states, then make Cypress tests reach them: open navigation menus and dialogs, expose validation messages, and exercise forms. Run checks after the relevant UI is in place, rather than treating one scan of the initial page as coverage for the entire application. Cypress describes its accessibility workflow and the distinction between automated checks and broader accessibility evaluation in its accessibility testing guide.
As an Amazon Associate I earn from qualifying purchases.
- Choose where scans run. For a code-first workflow, Cypress documents using the community
cypress-axeplugin: inject axe-core and callcheckA11y()on the page or component state. You can configure findings to fail the test. Scanning adds runtime, especially when repeated across many unique states. Cypress Accessibility instead processes scans in Cypress Cloud as tests are recorded; Cypress describes it as a paid premium solution. - Reach meaningful states. Use end-to-end or component tests to visit views and interact with menus, dialogs, and forms. Cypress Accessibility can use test activity as context for some custom rules.
- Add application-specific assertions. Assert expected keyboard behavior, accessible names, and content such as image alternative text. A general scanner cannot know every expectation specific to your product.
- Use an agent to investigate, not rubber-stamp. Ask it to inspect test evidence, explain a finding, propose a focused change, and rerun the relevant test.
- Review the experience yourself. Check that the control’s name and behavior match its purpose, that keyboard interaction works, and that the change has not disrupted surrounding structure. Include manual checks and assistive-technology testing appropriate to your product.
Choose between in-test scans and Cypress Accessibility
The practical choice is whether you want to manage scanning in the test code or use Cypress’s Cloud workflow. The options differ in where scans run and what context the rules can use; choose based on your test architecture, runtime budget, and need for agent-ready reports.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →| Approach | Where scans run | Trade-off to consider |
|---|---|---|
cypress-axe with axe-core |
In your Cypress test, against the current rendered page or component state | Requires plugin setup and adds scan time to test execution; you control which states are scanned and how findings affect test results. |
| Cypress Accessibility | In Cypress Cloud after tests are recorded | Cypress describes it as a paid premium solution. Its rules can use test interactions as additional context for some findings, and it provides agent-ready accessibility reports. |
Cypress says axe-core detects “up to 57%” of issues that would appear in a manual accessibility audit, attributing the figure to Deque; the Cypress page does not state the year. “Up to” is a ceiling, not a detection guarantee for your site, test suite, or audit. Cypress also cautions that automated checks cannot establish that an application is accessible, particularly for heavily custom interfaces or questions requiring human judgment. See How Cypress Accessibility works and its Cypress accessibility rules.
Write assertions for behavior scanners cannot infer
Rule-based scans are useful for detectable patterns, but application intent often needs a direct test. Cypress documents native Tab-event testing with cy.press() and demonstrates asserting expected image alternative text in its accessibility testing guide. Adapt assertions to the user-facing requirement rather than merely checking that an attribute exists.
- Keyboard flow: verify that Tab and other relevant keys move focus and activate controls as intended in the state under test.
- Accessible names: assert that important controls expose names that identify their purpose.
- Alternative text: for meaningful images, check the expected text; for decorative images, ensure they are not announced as meaningful content.
- State changes: after opening a dialog or submitting a form, test the resulting focus and feedback behavior your interface requires.
These checks encode your application’s expectations; they do not replace checking the interface with assistive technologies and human judgment.
Give an AI agent evidence it can inspect
Cypress documents a workflow combining Cypress skills for writing and reviewing tests, cypress tap for driving an open Cypress session and inspecting its command log, failures, DOM, and accessibility tree, and Cypress Cloud MCP or Cloud CLI for recorded failures and replay context. The agent can use the accessibility tree to identify landmarks, headings, forms, controls, roles, and accessible names before proposing assertions. Cypress outlines its AI capabilities in the AI overview, Cypress AI Skills, and cypress tap documentation.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Set up cypress tap with the documented constraints
The current cypress tap documentation specifies Cypress 15.21.0 or later, a live cypress open session, and a supported Chromium-based browser: Chrome, Chromium, Edge, or Electron. It does not attach to headless cypress run sessions. These are version-sensitive requirements, so confirm the current documentation when setting up the workflow.
- Start Cypress in open mode with a supported browser and load the application state you want the agent to inspect.
- Give the agent a focused task: identify the failed assertion or accessibility finding, inspect relevant DOM and accessibility-tree evidence, and propose a minimal change.
- Have it rerun the relevant test after the change and report the new result and any remaining findings.
- Review the change and test the interaction yourself before treating the issue as resolved.
Cypress says many AI capabilities—including Cypress skills, cypress tap, Cloud MCP, and Cloud CLI—work on the free Starter plan, while agent-ready accessibility reports require Cypress Accessibility. Plan packaging can change; check the current AI overview before making a purchasing decision.
Interpret findings without mistaking a scan for an audit
Axe Core catches common issues such as missing accessible names or image alternative text. Cypress Accessibility supplements the standard Axe Core rules with Cypress rules that use test interactions as context. For example, a clickable div or span used as a control may be better represented by a semantic button or link.
Automated results are evidence about the rules and states that were checked, not a certification of accessibility or compliance. Heavily custom interfaces and questions about whether an interaction makes sense to people require human review. The Cypress Team makes the same distinction in its April 21, 2026 article on Cypress Accessibility and AI agents: agents can propose solutions and trade-offs and help triage findings, but cannot independently solve all accessibility issues.
Troubleshoot common workflow failures
- The scan reports no findings, but a problem remains. Confirm that the test visited the affected state. A scan cannot inspect an unopened menu or unrendered validation message, and a clean automated result does not establish full accessibility.
- Tests are slower after adding scans. Scans take runtime, particularly when repeated across many unique states. Keep checks focused on meaningful states and consider whether the Cloud-based Cypress Accessibility workflow better fits your team.
- A clickable element triggers a semantic or interaction concern. Check whether a
divorspanis being used as a control. A semantic button or link may better communicate purpose and behavior to browsers and assistive technologies. - cypress tap cannot connect to the test. Confirm Cypress is 15.21.0 or later,
cypress openis running, and the session uses Chrome, Chromium, Edge, or Electron. The documented tool does not attach to headlesscypress run. - An agent suggests a change that passes the scan but feels wrong. Review the control’s purpose, accessible name, keyboard behavior, and surrounding structure; passing a rule is not enough to validate the experience.
Or skip the browser setup
For website screenshots outside this Cypress accessibility workflow, ScreenshotNeo is an API and MCP server. It is not a Cypress scanner or an accessibility audit tool. A single GET request can return a PNG, JPEG, WebP, or PDF; the API also has options for full-page captures and element-specific screenshots. See the ScreenshotNeo API documentation.
curl -G "https://api.screenshotneo.com/v1/shot" -d access_key=YOUR_API_KEY --data-urlencode url=https://stripe.com -o shot.webpScreenshotNeo accepts cookie and consent banners and removes more than 60 known consent platforms, newsletter popups, and chat widgets before capture; those steps can be turned off. Bot checks and CAPTCHAs, blank pages, timeouts, failed loads, and cache hits are not billed. Its MCP server provides take_screenshot, get_page_info, and capture_pdf for AI agents and other MCP clients. The free plan includes 1,000 shots per month with no card; paid plans start at $5 for 3,000 shots.
Sign up for ScreenshotNeo’s free plan to get 1,000 screenshots a month with no card.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minute
