Accessibility logo

Accessibility

Community
nahid-sparktales
accessibility

Build and fix interfaces so they meet WCAG 2.2 AA in practice — native semantics, keyboard paths, focus order and visibility, accessible names, contrast, form errors, live regions, target size. Fires while writing or reviewing UI code, when a component is keyboard- or screen-reader-hostile, or when an audit finding has to be turned into a change. Not for proving the result (accessibility-verification), not for viewport layout work (responsive-design), and not a recital of the standard.

Overview

Publishernahid-sparktales
Repositoryagent-dispatcher
Skill nameaccessibility
Stars
58
Forks
5
Bundled files
1
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by nahid-sparktales on GitHub. Read the source before you install it.

Installation

Install the Accessibility AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/nahid-sparktales/agent-dispatcher/tree/main/skills/design/accessibility
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/nahid-sparktales/agent-dispatcher.git /tmp/agent-dispatcher
mkdir -p .claude/skills
cp -r /tmp/agent-dispatcher/skills/design/accessibility .claude/skills/accessibility
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Accessibility in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Accessibility on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Accessibility is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Accessibility in practice

Most barriers are created by replacing something the platform already made accessible. The cheapest accessible component is the one you did not rebuild.

When this fires

While building or changing any interactive UI, and when turning an audit finding into a fix. It does not fire to prove a screen is accessible — that is accessibility-verification, and it needs a keyboard and a screen reader, not a code read.

Procedure

  1. Use the native element first. button, a[href], input, label, select, details, dialog, table. Each arrives with a role, keyboard behaviour, focus handling and state reporting you would otherwise have to write and maintain. A div with a click handler has none of it. If the project already has an accessible component for this, reuse that instead.
  2. Get the structure right before the styling. One h1 per view, headings in descending order with none skipped, real landmarks (header, nav, main, footer), lists marked up as lists, a unique page title per route. Screen reader users navigate by these; they are the table of contents, not decoration.
  3. Give every control a name that matches its visible label. Visible text is the name where possible; label[for] for inputs; an aria-label only for controls with no visible text, such as an icon button. When there is visible text, the accessible name must contain it — a button reading "Save" named "Submit form" breaks voice control.
  4. Walk the keyboard path yourself. Tab reaches everything interactive, in DOM order; Enter and Space operate it; Escape closes what opened; focus never gets stuck. Do not use positive tabindex — fix the DOM order instead. Composite widgets (menus, tabs, grids) take one Tab stop and move internally with arrow keys.
  5. Keep focus visible and unobscured. Never delete an outline without shipping a replacement. Style :focus-visible so it survives sticky headers, overflow clipping and dark backgrounds, and check that a sticky bar or cookie banner does not cover the focused element.
  6. Move focus deliberately when the view changes. Opening a dialog moves focus into it and traps it there; closing returns focus to the element that opened it; a client-side route change moves focus to the new heading or main region. Focus left behind on a removed node goes to body, and the user loses their place silently.
  7. Choose colours that carry the contrast. 4.5:1 for body text, 3:1 for large text and for the boundaries of controls, focus rings and meaningful graphics. Check hover, active, selected and error states too, and text sitting over images or gradients. Never let colour alone carry meaning — pair it with text, an icon or a pattern.
  8. Make forms explain themselves programmatically. Associate the error with its field (aria-describedby), mark the field invalid (aria-invalid), and write error text that says what is wrong and how to fix it. Use autocomplete tokens on personal-data fields. Do not make people re-enter information they already gave you in the same flow, and do not block paste in password or one-time-code fields.
  9. Announce what changes without a page load. A status message needs a live region that already exists in the DOM before it updates — aria-live="polite" for status, role="alert" for errors. Announce once, keep it short, and do not wire a live region to something that updates constantly.
  10. Respect the user's settings. Honour prefers-reduced-motion by removing movement, not by speeding it up. Respect prefers-color-scheme and prefers-reduced-transparency where the design uses them. Never disable zoom in the viewport meta tag.
  11. Cover the WCAG 2.2 additions that bite hardest. Any drag interaction needs a single-pointer alternative (click or button). Interactive targets are at least 24×24 CSS px unless spaced or inline in a sentence. Help links stay in the same relative place across pages.
  12. Add ARIA last, and as little as possible. No ARIA beats wrong ARIA: an incorrect role silently overrides the real one. Never put aria-hidden on anything focusable, and never invent role/state combinations — copy an established pattern and keep its keyboard contract intact.
  13. Stop and ask before changing shared surfaces. A contrast fix that alters a brand token, or a markup change inside a shared design-system component, affects screens you are not looking at. Propose the change with the reason; do not push it through on your own authority.

Checklist

  • Every interactive element is a native control or reuses an existing accessible component
  • Heading order and landmarks make sense read alone, with no skipped levels
  • Every control has a name, and it contains the visible label text
  • Tab reaches everything, in a sensible order, with no trap and no positive tabindex
  • Focus is always visible and never covered by sticky or overlaying chrome
  • Dialog and route changes move focus, and closing returns it
  • Text and control boundaries meet contrast in every state, including over imagery
  • Errors are associated with their fields and readable without colour
  • Dynamic messages land in a live region that existed beforehand
  • Reduced motion honoured; zoom not disabled
  • Drag interactions have a click alternative; targets are 24×24 CSS px or spaced
  • ARIA added only where a native element could not do the job

Failure handling

  • The component is a third-party widget you cannot change — say so, record the specific barrier, and look for a documented accessible mode or a replacement. Do not paper over it with ARIA that describes behaviour the widget does not have.
  • The fix needs a design decision (a token fails contrast, a layout has no room for a 24px target) — name the conflict and the options, and let the design owner choose. Do not quietly ship a visual change to satisfy a checker.
  • A pattern has no obvious accessible form — look up the established pattern for it rather than inventing roles. If the project pins a component library, check that library's current accessibility API before assuming what it supports.
  • A checker passes but the interaction still feels wrong — trust the interaction. Automated rules cannot see whether a name is meaningful or an order is logical.

Evidence to report

Name the elements changed and why: which control became a button, which name was added, where focus now moves, which token changed and its measured ratio before and after. State plainly that these are implementation changes — they are unverified until someone drives the keyboard and a screen reader. Hand the claim to accessibility-verification; do not call it conformant here.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Accessibility AI skill do?

Build and fix interfaces so they meet WCAG 2.2 AA in practice — native semantics, keyboard paths, focus order and visibility, accessible names, contrast, form errors, live regions, target size. Fires while writing or reviewing UI code, when a component is keyboard- or screen-reader-hostile, or when an audit finding has to be turned into a change. Not for proving the result (accessibility-verification), not for viewport layout work (responsive-design), and not a recital of the standard.

Why use Accessibility on TypingMind?

Because you install it once and use it with any model. Accessibility is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Accessibility in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/nahid-sparktales/agent-dispatcher/tree/main/skills/design/accessibility. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Accessibility?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Accessibility?

As many as you like. As long as a model supports skills, you can use Accessibility with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Accessibility AI skill free?

Yes. It is published on GitHub by nahid-sparktales under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇