Slide Accessibility logo

Slide Accessibility

Community
andreworia
slide-accessibility

Makes a deck usable with a screen reader and compliant for a regulated client — alt text, reading order, slide titles, contrast, table headers, captions. Not for general visual polish.

Overview

Publisherandreworia
Repositoryclaude-pptx-skills
Skill nameslide-accessibility
Stars
4
Forks
2
Bundled files
Instructions only
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.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by andreworia on GitHub. Read the source before you install it.

Installation

Install the Slide 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/andreworia/claude-pptx-skills/tree/main/skills/slide-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/andreworia/claude-pptx-skills.git /tmp/claude-pptx-skills
mkdir -p .claude/skills
cp -r /tmp/claude-pptx-skills/skills/slide-accessibility .claude/skills/slide-accessibility
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Slide 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 Slide 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 Slide 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.

Slide Accessibility

When to use

Use this skill when a deck goes to a public body, a bank, a healthcare or education client, anyone procuring under an accessibility standard, or any audience that includes a screen-reader user. Also use it before a deck is published to a website, because a published .pptx or its PDF inherits every defect. Do not use it as a general design pass; contrast and reading order are correctness problems, not taste problems, and brand-compliance-auditor covers the rest.

What it does

It runs the deck against the checkable accessibility requirements, fixes them in the file rather than in a separate document, and produces a defect list with what was changed. The output is a remediated deck plus an audit trail a procurement team can accept.

Method

1. Know which standard you are held to. In the United States, Section 508 incorporates WCAG 2.0 Level AA; in Europe, EN 301 549 references WCAG 2.1 Level AA and underpins public-sector procurement. Most contracts name WCAG AA. Ask which, and write the answer at the top of the audit.

2. Give every slide a unique title. It is the primary navigation handle: a screen reader announces it, and Outline view and slide-jump tools use it. Every slide needs one and no two should be identical — three slides called "Agenda" defeat the point. Where the design carries no visible title, hide the placeholder rather than deleting it: with the Accessibility Checker open, the Accessibility tab's Slide Title menu offers Add Slide Title and Hidden Slide Title, which puts a real title off the visible canvas.

3. Write alt text that carries the information, and mark the rest decorative. Right-click an object and choose Edit Alt Text, or use the Accessibility tab. What matters:

  • Describe what the image means here, not what it depicts. A revenue chart's alt text is the finding — "Revenue grew from £12m to £31m over four years, with the step change in FY23" — not "bar chart". Do not open with "image of"; the screen reader already says so.
  • Decorative rules, textures and brand furniture get the Mark as decorative checkbox, which removes them from the reading order. That matters as much as describing content.
  • A screenshot of a table is invisible. Rebuild it as a real table.
  • Alt text on a group is read and its members are skipped, so give a composite diagram one good group-level description rather than eleven fragments.

4. Fix the reading order, and know which pane you are in. Two tools, and they run in opposite directions:

  • Reading Order Pane (Review > Check Accessibility, then the Reading Order Pane button on the Accessibility tab) lists objects top to bottom in reading order — the top item is read first. Windows desktop only; it is not in PowerPoint for the web or on macOS.
  • Selection Pane (Home > Arrange > Selection Pane) lists the same objects in z-order, which is the reverse — the bottom item is read first. This is the single most common remediation mistake: a deck "fixed" in the Selection Pane top-down comes out backwards.

Order should follow the argument: title, then exhibit or body, then source line and footnotes. Re-check after any layout change, because a newly inserted shape lands at the front.

5. Check contrast against the real numbers. WCAG AA requires 4.5:1 for normal text and 3:1 for large text, where large means 18 pt or larger, or 14 pt bold or larger. Meaning-bearing non-text elements — chart series, an icon that is the only label — need 3:1. The built-in checker flags some contrast problems but misses text over a photograph, a gradient or a semi-transparent panel, so sample the actual pixel behind the text. Where a brand colour fails, do not invent one: use the palette's approved dark tint, or put the text on a solid panel.

6. Never encode meaning in colour alone. A red bar and a green bar mean nothing to a colour-blind reader or in greyscale print. Add a direct label, pattern or symbol alongside the colour, and test by exporting a slide in greyscale.

7. Build tables as tables. A grid of aligned text boxes looks like a table and exports as loose text. Use a real table object, tick Header Row in Table Design > Table Style Options, and keep it rectangular: no merged or split cells, no blank rows used as spacing, no nested tables. PowerPoint tables cannot express column and row scope, so a table that needs both is a sign it should be split into two.

8. Handle media and motion. Video needs captions: select it and use Playback > Insert Captions, which accepts a WebVTT (.vtt) file. Audio needs a transcript, which can live in the notes. Nothing may flash more than three times per second. Avoid autoplay, and avoid animations that carry information the static slide does not — a build that reveals the punchline is lost in the exported file.

9. Make links say where they go. "Click here" and a bare URL are both useless read aloud. In Insert > Link, put a meaningful phrase in Text to display and add a ScreenTip where the destination is not obvious.

10. Run the checker, then export correctly. Review > Check Accessibility sorts findings into Errors, Warnings and Tips; work Errors first and tick Keep accessibility checker running while I work. It is a floor, not a ceiling — it cannot judge whether alt text is good. On export, leave Document structure tags for accessibility on (see pptx-export-and-pdf); an untagged PDF discards everything above.

Inputs

  • The deck, and the standard being applied.
  • The brand palette with hex values, so contrast is computed against approved colours.
  • Whether the deliverable is the .pptx, a PDF, or both.
  • Any embedded media, and whether captions or transcripts exist.

Output format

  • The standard applied, stated at the top.
  • A defect list by slide: issue, severity, and the fix made.
  • Contrast as measured ratios against the 4.5:1 and 3:1 thresholds, naming failing pairs.
  • Reading order changes, with the pane used.
  • Anything that could not be remediated in PowerPoint, and why.
  • The checker's final state and the export settings used.

Example

Input: a 24-slide programme update for a public-sector client, contracted to WCAG 2.1 AA.

Output:

  • Standard: WCAG 2.1 AA (EN 301 549).
  • Titles: 24 of 24 titled; slides 3, 11 and 20 use hidden titles because the design is full-bleed. Three duplicate "Progress" titles disambiguated.
  • Alt text: 31 objects described, 14 decorative motifs marked decorative, chart alt text on slides 6, 9 and 14 rewritten to state the finding.
  • Reading order: corrected on 9 slides in the Reading Order Pane. Slide 17 had been "fixed" bottom-up previously and was fully reversed.
  • Contrast: brand mid-grey #8A8F98 on white measured 2.9:1 at 12 pt and fails; replaced with #5A6069 at 4.9:1. Amber chart series at 2.1:1 darkened and direct-labelled.
  • Colour alone: RAG status on slides 8 and 15 given text labels.
  • Tables: slide 12's text-box grid rebuilt as a real table with Header Row; slide 19's merged header cells split.
  • Media: the embedded video captioned from a supplied .vtt.
  • Export: PDF with structure tags on. Checker reports zero Errors, two Tips retained deliberately.

Frequently asked questions

What does the Slide Accessibility AI skill do?

Makes a deck usable with a screen reader and compliant for a regulated client — alt text, reading order, slide titles, contrast, table headers, captions. Not for general visual polish.

Why use Slide Accessibility on TypingMind?

Because you install it once and use it with any model. Slide 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 Slide Accessibility in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/andreworia/claude-pptx-skills/tree/main/skills/slide-accessibility. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Slide 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 Slide Accessibility?

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

Is the Slide Accessibility AI skill free?

Yes. It is published on GitHub by andreworia 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 👇