Skip to content

ebad66/a11y.sense

Repository files navigation

a11y.sense

a11y.sense landing page

"Something like 99% of all websites have accessibility issues. Assume there's something wrong and you're probably right." — Disabled Reddit user

1.3 billion people live with a disability. Most of the web wasn't built for them. a11y.sense fixes that.


What is a11y.sense?

a11y.sense is an AI-powered accessibility QA sandbox that audits any live URL against the Web Content Accessibility Guidelines (WCAG) — and actually tells you how to fix what's broken.

Paste a URL. Get a full accessibility audit mapped to WCAG principles. See every issue visualized on an interactive 3D human skeleton where each body system represents a disability category. Click the brain for cognitive issues. Click the hands for motor issues. No jargon. No guesswork.


The Visualizer

a11y.sense visualizer — Perceivable view

Every issue is mapped to the body system it affects. Perceivable lights up the skull and sensory anatomy. Operable targets the hands. Understandable hits the brain. Robust traces the nervous system. Click any body part to drill into the exact issues, the affected users, and the fix.

a11y.sense visualizer — Robust / Nervous System view

Each issue panel shows you:

  • The WCAG principle and success criterion it violates
  • Who is affected and how many people globally
  • A plain-English description of why it matters
  • A suggested fix with the exact CSS class or element to target
  • The raw affected HTML element — ready to paste into your PR

Developer Handoff

Dev Handoff — markdown export

When you're done auditing, generate a developer-ready .md report per WCAG principle. Copy it straight into a GitHub issue or download it for your PR. Zero friction from audit to fix.



Summarized PDF

PDF Download

Download and share a .pdf with team members or clients to easily discuss suggested changes. The pdf also includes estimated effort, suggested owner, and ROI for each issue; making prioritization simpler.


Features

  • Live URL auditing — paste any URL, get real results in seconds
  • 5 parallel AI audit profiles — running simultaneously across all WCAG core principles
  • WCAG-mapped issue reporting — every issue tagged with the exact guideline it violates
  • Actionable fixes — not just what's broken, but how to fix it with precise CSS selectors
  • Interactive 3D skeleton — WCAG's four abstract principles made immediately human and intuitive
  • Gemini Vision screenshot analysis — visual layer inspection on top of DOM auditing
  • Export PDF Summary — easily generate a .pdf summary and analysis of detected issues
  • Playwright-powered element resolution — finds exact coordinates and selectors for every flagged element
  • Dev Handoff export — one-click .md report per principle, ready for GitHub

How It Works

Your URL
   │
   ▼
Playwright scrapes the live page → extracts DOM, screenshots, element coordinates
   │
   ▼
Gemini Flash runs 5 parallel audits across all WCAG profiles simultaneously
   │
   ▼
Structured JSON: issues, WCAG codes, CSS selectors, severity, fixes
   │
   ▼
3D skeleton maps each issue to the body system and disability it affects
   │
   ▼
You ship accessible code

Tech Stack

Layer Tech
Frontend Next.js + React 19
3D Visualization Three.js
AI Auditing Google Gemini Flash
Visual Analysis Gemini Vision
Browser Automation Playwright

WCAG Principles → Skeleton Mapping

WCAG Principle Body System Linked Disabilities
Perceivable Eyes / Ears / Skull Visual, Auditory
Operable Hands / Limbs Motor
Understandable Brain Cognitive, Neurological
Robust Nervous System Assistive technology users
Navigation Hands / Limbs Motors

Why We Built This

The tools that exist today either overwhelm developers with raw WCAG text or produce surface-level reports with no path to a fix. Neither helps. We built a11y.sense because accessibility shouldn't require a specialist — it should be obvious, immediate, and baked into the workflow.

The Playwright → Gemini pipeline we built is genuinely different. It doesn't just scan static HTML. It loads the page, resolves elements in context, and audits what a real user actually encounters. Only 3% of websites meet full a11y standards. That number should be embarrassing. We're here to change it.


What We Learned

Building a11y.sense gave us a deep appreciation for the craft of AI prompt engineering. Getting Gemini to return consistent, structured JSON across five parallel audit profiles taught us that the difference between a useful AI output and an unusable one often comes down to how precisely you constrain the prompt. We also came face-to-face with just how pervasive accessibility gaps are in the real world — auditing live URLs made it viscerally clear that accessibility is still treated as an afterthought by the vast majority of developers, reinforcing why tools like a11y.sense genuinely matter.


What's Next

  • Mobile app accessibility — bringing WCAG-mapped auditing to iOS and Android
  • IDE plugin + CI/CD integration — catch accessibility issues at commit, not after deploy

Built By

GitHub LinkedIn
Ebad Ahmad github.com/ebad66 linkedin.com/in/ebadahmad
Aayan Atif github.com/aayanA6 linkedin.com/in/aayanatif
Shayan Syed github.com/Shay350 linkedin.com/in/shayansyed1

Made with conviction at GenAIGenesis.


License

MIT

About

1.3 billion people live with a disability. a11ysense is an AI-powered sandbox that simulates blind, deaf, motor-impaired, and cognitive user experiences on any live URL. Built at GenAIGenesis 2026.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors