More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Accessible by Default: ARIA, Keyboard, and Contrast

    1

    Make any web page usable by keyboard and screen reader first: a three-pass audit order, native-first markup, focus contracts for dialogs, menus, tabs, and forms, and a scored gate.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Accessible by Default: ARIA, Keyboard, and Contrast

    Accessible by Default: ARIA, Keyboard, and Contrast

    Example session with this skill installed

    I run a fictional community theater called Riverside Playhouse. Our ticket page needs an accessibility review before Saturday's show. The main task is booking two tickets. Run your three passes and give me the findings log plus the gate score.

    • Read your context and instructions
    • Compiled the accessible by default

    AUDIT — Riverside Playhouse ticket page. Invented scenario and numbers for this demo only.

    Main task: book two tickets for Saturday's show.

    Keyboard pass, 4 barriers: the seat picker was clickable containers, unreachable by keyboard — rebuilt as real buttons. The first fix tried role="button" on the containers without key handling; the skill's native-first rule caught it and the buttons replaced them. The date selector trapped Tab inside it — focus now moves through and out. Stepper buttons had no visible focus — strong ring restored. The promo-code dialog ignored Escape — now closes and returns focus to its trigger.

    Screen-reader pass, 3 barriers: two top-level headings with a skipped level — outline rebuilt to one top heading with ordered sections. The seat map announced only as an image — now a text summary of sections and available seats, plus a list alternative. "Tickets added" never announced — now in a polite live region.

    Visual pass, 2 barriers: available-seat text measured 2.8 to 1 — darkened to 4.7 to 1. Sold-out seats shown by color alone — now labeled "sold out" in text.

    First gate score: 10 of 12. Images and live updates scored 1 — the seat summary named sections but not rows. Visual scored 1 — the legend overlapped at 200% zoom. Both fixed: the summary now names rows, the legend stacks. Retest: 12 of 12. The page ships.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Most accessibility work starts with labels and ends there — and the page still locks people out. This skill starts where the barriers actually are: the keyboard. It runs a fixed three-pass audit — keyboard first, screen reader second, visual last — because a page that works with a keyboard alone is most of the way there, and the later passes check structure read aloud, then contrast, scale, and motion. It gives you the native-first rule (use the element the browser provides before reaching for extra attributes), focus contracts for dialogs, disclosures, tabs, forms, and custom menus, a findings-log template that records barrier, who it blocks, fix, and retest, and a six-line scoring rubric: a page ships when it reaches 10 from a possible 12, and no line may score 0. Edge cases cover pages that change without reloading, widgets you cannot rewrite, data tables, charts, video, and time limits; the failure-mode catalog names the attributes-first fix, the checker-only verdict, the invisible focus ring, placeholder-as-label, the container button, color-only errors, and the one-time audit.

    • It is not a legal-compliance verdict or a certification; it produces evidence that the main task works for keyboard users, screen-reader users, and low-vision users.
    • It cannot replace testing with people who use assistive technology; it finds and fixes the barriers you can find yourself so that testing is spent on real use.
    • It cannot fix a third-party widget you do not control; it labels what you can and asks for an alternative path to the same task.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 3 days ago

    Permissions

    No special permissions declared or detected

    Listed3 days ago

    What's inside

    Frequently Asked Questions

    No. It finds the barriers you can find yourself and fixes them first, so a testing session is spent on real use rather than basics the skill would have caught.

    Automated checks catch about a third of issues. This skill adds the keyboard and screen-reader passes — the test itself — plus the component contracts and the scored gate.

    Label what you can and always provide an alternative path to the same task — a phone number, a plain form, a link. An unlabeled widget with no alternative is a dead end.