neura11y

Accessibility knowledge, structured for AI.

Neura11y

neura11y is a structured accessibility knowledge base for AI context loading, RAG pipelines, and accessibility-aware content generation across web, documents, mobile apps, social media, screen-reader workflows, voice UI, and physical ICT.

neura11y.com

v0.3.0 • standards • workflow guides • prompt packs • validation tooling

neura11y packages normative standards, applied workflow guidance, prompt templates, fetched provenance captures, and validation tooling into a single repository so AI systems can produce accessibility guidance and accessible outputs with stronger traceability and less hallucination.

neura11y enables AI systems to:

• Generate accessible HTML, content, and documents without hallucinating requirements
• Generate or review accessibility guidance for native mobile apps and screen-reader test workflows
• Cite specific WCAG success criteria with correct thresholds and level designations
• Apply platform-specific social media accessibility conventions
• Distinguish normative requirements from prescriptive best practices
• Produce legally defensible guidance grounded in Section 508, WCAG, and the EU EAA

Read AI-USAGE-GUIDE.md first. It defines context-loading patterns for different task types.

Quick navigation: Use INDEX.md to find files by standard, domain, or tag without loading the full repository.

For AI pipelines: Chunk at ## heading level. Preserve source_file, chunk_id, title, heading_path, standard, sc_number, domain, tags, platform, status, type, source_url, content_family, content_kind, level, last_fetched, last_verified, last_reviewed, and stale as chunk metadata.

Canonical files: Prefer the curated and split canonical files referenced by INDEX.md and AI-USAGE-GUIDE.md, especially for Core-AAM, HTML-AAM, and WCAG2ICT. Treat *-fetched.md files as raw captures and *-full-fetched.md files as raw full provenance, not the default context payload.

This repository includes a /a11y Claude Code slash command at .claude/commands/a11y.md. When Claude Code is open in this directory, type /a11y followed by your task and it will automatically load the right knowledge files and produce grounded guidance with proper SC citations.

/a11y audit this button: <button onclick="go()">Click</button>
/a11y generate an accessible modal dialog in React
/a11y check our Instagram post for accessibility issues
/a11y does our procurement process need to comply with Section 508?
/a11y test with JAWS on our checkout flow

The skill routes across 12 patterns (web audit, component generation, social media, documents, legal compliance, voice UI, kiosk/embedded ICT, and 4 screen-reader tools), reads only the files relevant to the matched pattern, and enforces output rules: SC citations in SC X.X.X — [Title] (Level A|AA|AAA) format, [NORMATIVE]/[PRESCRIPTIVE] prefixes, [FAIL]/[WARN]/[BEST PRACTICE] severity markers, and a manual testing checklist on every response.

claude-a11y-repo/
├── VERSION                   — Repository release version
├── README.md                 — This file
├── INDEX.md                  — Machine-readable cross-reference map
├── GLOSSARY.md               — Canonical definitions for all key terms
├── AI-USAGE-GUIDE.md         — Context-loading patterns for AI systems
├── index.html                — Project landing page
├── logos/                    — Branding assets
├── .claude/commands/         — Claude Code slash commands (e.g. /a11y)
│
├── meta/                     — Repository management metadata
│   ├── standards-registry.md — Master list of all standards with source URLs
│   ├── fetch-log.md          — Record of auto-fetch operations
│   ├── fetch-log.jsonl       — Structured fetch history for tooling
│   ├── ai-validation-fixtures.json — Validation fixtures for AI usage patterns
│   ├── chunk-manifest.jsonl  — Exported AI-ingestion corpus
│   ├── chunk-manifest-summary.json — Chunk-export summary statistics
│   ├── freshness-manifest.json — Per-file freshness signals for tooling
│   ├── latest-ai-validation-report.md — Latest validation run report
│   ├── update-schedule.md    — Maintenance cadence per content type
│   └── coverage-gaps.md      — Known gaps and planned additions
│
├── standards/                — Normative standards content (Layer 1)
│   ├── wcag/                 — WCAG 2.1, 2.2, 3.0
│   ├── aria/                 — WAI-ARIA 1.2, APG, ARIA in HTML
│   ├── section-508/          — US Section 508 technical standards
│   ├── pdf-ua/               — PDF/UA (ISO 14289)
│   ├── epub/                 — EPUB Accessibility 1.1
│   ├── en-301-549/           — European standard EN 301 549
│   └── other-standards/      — ATAG 2.0, UAAG 2.0, ISO 9241-171
│
├── domains/                  — Applied guidance by content domain (Layer 2)
│   ├── web/                  — HTML, ARIA, components, testing
│   ├── documents/            — Word, PowerPoint, PDF, Excel, plain language, email, Google Workspace
│   ├── mobile/               — Native iOS and Android app guidance
│   ├── social-media/         — Alt text, captions, 7 platform guides
│   ├── voice/                — Voice UI and speech-command guidance
│   └── physical-ict/         — Kiosk and embedded system playbooks
│
├── cognitive/                — COGA, cognitive accessibility patterns
├── color-and-visual/         — Contrast ratios, APCA, typography
├── media/                    — Images, video, audio accessibility
├── screen-readers/           — AT guides, workflows, support matrices, and quirks
│   └── source/               — Fetched vendor source docs and keystroke references
├── ai-prompts/               — Ready-to-use prompt templates (Layer 3)
├── legal-and-compliance/     — ADA, Section 508, EU EAA, UK, Canada, Australia
├── reference/                — Tools, statistics, citations
└── scripts/                  — Fetch, validation, chunk-export, and maintenance scripts

• WCAG 2.1 (W3C Recommendation, June 2018) — 78 success criteria
• WCAG (current stable W3C Recommendation: 2.2, October 2023) — 86 active success criteria in the current version (9 new; SC 4.1.1 obsolete)
• WCAG 3.0 (Working Draft) — Overview and emerging guidance
• WAI-ARIA 1.2 (W3C Recommendation, June 2023)
• ARIA in HTML (W3C Recommendation)
• Accessible Name and Description Computation 1.2 (W3C Working Draft, March 2026)
• Core Accessibility API Mappings 1.2 (W3C Candidate Recommendation Draft, March 2026)
• HTML Accessibility API Mappings 1.0 (W3C Working Draft, March 2026)
• ARIA Authoring Practices Guide (APG)
• WCAG2ICT 2.2 (W3C Group Note, December 2025)
• Section 508 (US, 2017 refresh)
• PDF/UA (ISO 14289-1:2014)
• EPUB Accessibility 1.1 (W3C Recommendation, May 2023)
• EN 301 549 (European standard, v3.2.1)
• ATAG 2.0, UAAG 2.0
• ISO 9241-171 (software accessibility overview)

• Web — HTML semantics, ARIA, focus management, keyboard navigation, forms, components, SPAs, mobile patterns, PWAs, testing
• Documents — Microsoft Word, PowerPoint, PDF, Excel; federal plain language guidelines; PDF form and Excel data-viz examples
• Document Workflows — Accessible email/newsletters and Google Docs/Slides/Sheets authoring
• Mobile — Native iOS and Android accessibility patterns for labels, focus, gestures, scaling, and announcements
• Social Media — Twitter/X, LinkedIn, Instagram, Facebook, TikTok, YouTube, Mastodon; sign-language and terminology notes
• Screen Reader Testing — Cross-tool workflows, mobile screen-reader testing, support matrices, quirks, and JAWS/NVDA/VoiceOver/TalkBack guides
• Voice — Voice command and spoken UI accessibility patterns
• Physical ICT — Kiosk and embedded system accessibility playbook
• Production Workflows — CMS authoring workflow and design-to-development handoff guidance

• Release 0.3.0: Adds fetchable AccName, Core-AAM, HTML-AAM, WCAG2ICT, WAI tutorials, and screen-reader vendor sources; splits oversized standards into AI-usable canonical files with raw full captures preserved for provenance.
• Screen-reader workflows: Coverage now includes cross-tool workflow guidance, mobile screen-reader testing, common announcement and quirk references, fetched vendor source docs under screen-readers/source/, and four dedicated screen-reader prompt templates.
• Tooling references: The repo now includes an AI-first checker and CLI routing file at reference/accessibility-checkers-and-cli-tools.md, alongside the broader tools catalog plus automated and manual testing guidance.
• AI operations: The repository now includes schema validation, reference auditing, enriched chunk export, freshness sync, release-metadata sync, and fixture-based AI validation scripts under scripts/, with generated outputs in meta/chunk-manifest.jsonl, meta/chunk-manifest-summary.json, meta/freshness-manifest.json, and meta/latest-ai-validation-report.md.
• Prompt coverage: Prompt templates now cover the 8 core context-loading patterns in AI-USAGE-GUIDE.md plus 4 screen-reader workflows; the latest validation report passes structural checks and all 12 pattern preflight checks.
• Remaining backlog: The primary optional backlog item remains video game accessibility, which is still out of scope for v1.

Auto-fetchable standards and vendor source docs are refreshed via scripts in /scripts/. For a normal refresh or release pass, run scripts/fetch-all.py and then scripts/refresh-repo.py. Use scripts/export-ai-context.py for direct JSONL regeneration, scripts/sync-freshness.py to update registry freshness signals, scripts/validate-frontmatter.py and scripts/validate-references.py for structural checks, and scripts/validate-ai-suite.py for fixture validation (or response validation with --responses-dir). See meta/update-schedule.md for cadence.

See meta/coverage-gaps.md for known gaps and priority areas. All files must follow the frontmatter format defined in AI-USAGE-GUIDE.md.