Skip to content

Guide · Agent skill

How to use the Screen Reader Testing agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The screen-reader-testing skill gives an AI agent a structured guide for checking how a website or web app sounds to people who use screen readers, the programs that read interface content aloud. It is for front-end developers, QA testers and accessibility reviewers who want a repeatable way to test with VoiceOver, NVDA, JAWS, TalkBack or Narrator. After this guide you will know how to install the skill, how to ask for a test in plain language, and what the agent will check.

Screen Reader Testingwshobson/agents · MIT

Quick start

  1. 1Download the ZIP. Download the skill ZIP file from this page.
  2. 2Unzip into a skills folder. Unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder for one project.
  3. 3Or upload it in the Claude apps. In the Claude apps, open Settings, go to Capabilities, then Skills, and upload the ZIP.
  4. 4Install your screen readers. Install the screen reader software you will test with, such as VoiceOver, NVDA, JAWS, TalkBack or Narrator. The skill does not include them.
  5. 5Ask for a test in plain language. Describe the page or component, name the screen reader and browser, and say what to check, such as a modal or a form.
  6. 6Run the checks and review results. Follow the agent's checklist with the real screen reader, then confirm each announcement by listening, not by assuming.

What the screen-reader-testing skill makes the agent do

Once installed, the agent follows a guide for testing web applications with screen readers. It explains which screen reader and browser pairs to test, how to move between browse (virtual) mode for reading and focus or forms mode for filling in controls, and what to check in modals, live announcements, tabs and forms.

The skill also suggests ARIA markup patterns and includes a small JavaScript helper that logs what a screen reader would see for an element, such as its role, name and state. It is testing guidance, not a conformance verdict.

  • Pairs to test, from a minimum set (NVDA with Firefox, VoiceOver with Safari on macOS and iOS) to a comprehensive set that adds JAWS, TalkBack and Narrator
  • Reading and form-filling modes
  • Modal dialogs, live regions, tab interfaces and forms
  • ARIA patterns and a logging helper for debugging

Install the screen-reader-testing skill

Download the ZIP file for the skill from this page. Then either unzip it into your personal skills folder at ~/.claude/skills/, or into a project's .claude/skills/ folder if you want the skill available only in that project.

You can also upload the skill in the Claude apps. Go to Settings, then Capabilities, then Skills, and upload the ZIP there.

  • Personal install: unzip into ~/.claude/skills/
  • Project install: unzip into the project's .claude/skills/ folder
  • Claude apps: Settings → Capabilities → Skills

Install the screen readers you plan to test with

The skill does not include screen reader software. You install each one yourself. The skill's reference files cover setup for VoiceOver on macOS and iOS, NVDA and JAWS on Windows, and TalkBack on Android. You will also need a web browser such as Safari, Firefox, Chrome or Edge.

For NVDA, the reference file gives the download source as nvaccess.org and lists Ctrl + Alt + N as a start shortcut when the Windows shortcut is enabled. For VoiceOver on iOS, enable it in Settings → Accessibility → VoiceOver.

  • VoiceOver: macOS and iOS, used with Safari
  • NVDA: Windows, used with Firefox or Chrome
  • JAWS: Windows, used with Chrome
  • TalkBack: Android, used with Chrome
  • Narrator: Windows, used with Edge

Ask for a test with natural requests

You do not need special syntax. Describe the page or feature and the screen reader you care about, and mention what you want checked. The skill's description says it applies when you are validating screen reader compatibility, debugging accessibility issues, or ensuring assistive technology support.

Name the screen reader and browser if you have a preference. Name the kind of component too, such as a modal, a form, a tab set or a live status message, so the agent uses the matching part of the guide.

  • Test this checkout form for screen reader compatibility, using NVDA with Firefox.
  • Check whether the delete confirmation modal traps focus and returns it to the trigger button when VoiceOver is used in Safari.
  • Review this tab interface for ARIA and keyboard behaviour before we merge.
  • Why does our upload progress message not get announced? Help me debug it.
  • Give me a VoiceOver checklist for iOS Safari on this login page.

Check the common scenarios the skill covers

The skill walks through three common patterns. For modal dialogs, it shows a role=dialog structure with aria-modal and labelling attributes, plus focus moved into the dialog, focus trapped with Tab, Escape to close and focus returned on close.

For live regions, it compares polite status messages, assertive alerts, progress bars and logs that announce only additions. For tab interfaces, it shows the tablist and tabpanel structure and arrow, Home and End key handling.

  • Modal dialogs: focus entry, focus trap and focus return
  • Live regions: polite versus assertive announcements
  • Tabs: roles, selected state and arrow-key navigation
  • Forms: labels, errors and form-field navigation
<div role="status" aria-live="polite" aria-atomic="true"></div>

Use the debugging helper and the safety notes

The skill includes a JavaScript helper named logAccessibleName. It takes an element and logs its role or tag name, a name taken from aria-label, aria-labelledby or text content, its expanded, selected, checked and disabled state, and whether it is visible. It is a rough aid for inspecting markup, not a replacement for a real screen reader pass.

The skill does not run scripts and does not use the network, according to its listing. It does not install screen reader software for you, and it does not replace testing with real disabled users. The guide itself says to test with actual screen readers rather than simulators.

  • Runs scripts: no
  • Uses network: no
  • Screen reader software is installed separately by you
  • Helper output is a starting point, not a final result
console.log(logAccessibleName(document.querySelector('button')));

What people use it for

  • Checking a new modal dialog for focus entry, focus trapping and focus return before release
  • Confirming that form errors and status messages are announced in NVDA with Firefox
  • Running a VoiceOver pass on iOS Safari for headings, links and form controls using the rotor
  • Reviewing a tab interface for ARIA roles and arrow-key behaviour
  • Debugging why a progress or alert message is silent or interrupts speech

Frequently asked questions

Is the screen-reader-testing skill free?

The skill is listed under the MIT licence. The screen reader software it refers to, such as NVDA, JAWS or VoiceOver, is installed separately, and the skill's listing does not state pricing for those tools.

What do I need to use this skill?

You need screen reader software such as VoiceOver, NVDA, JAWS, TalkBack or Narrator, installed by you, and a web browser such as Safari, Firefox, Chrome or Edge. The skill itself does not run scripts or use the network.

How do I install the screen reader testing skill?

Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder. You can also upload it under Settings, then Capabilities, then Skills in the Claude apps.

Can the skill replace testing with real screen reader users?

No. The guide itself says to test with actual screen readers, not simulators, and it does not replace feedback from disabled users.

Does this skill certify that my site meets accessibility law?

No. The skill gives testing guidance only and does not issue a conformance or compliance verdict.

Which screen reader and browser should I test first?

The skill's minimum coverage is NVDA with Firefox on Windows, VoiceOver with Safari on macOS, and VoiceOver with Safari on iOS. It suggests adding JAWS with Chrome, TalkBack with Chrome and Narrator with Edge for more complete coverage.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use this skill if you are a developer, QA tester or accessibility reviewer who wants a structured checklist for VoiceOver, NVDA, JAWS or TalkBack passes on modals, live regions, tabs and forms. Skip it if you need legal compliance certification, testing of native mobile UI beyond the basic iOS and Android screen reader guidance, or a substitute for real disabled users testing your product.

Download the skill