Building an automated WCAG accessibility scanner

About this project

Manual WCAG audits are slow, expensive, and inconsistent. The tooling that exists is either too technical for non-specialists or too shallow to be actionable.

I built this tool to address a pain point I experienced directly while running accessibility audits for clients: getting from a raw list of violations to a clear, prioritised remediation plan takes significant effort. The scanner automates the first half of that process.

What we were up against

Most accessibility testing tools output raw violation data. That's useful if you're a developer who already knows what "aria-required-attr" means. It's not useful if you're a product manager trying to brief an engineering team on what to fix and why.

The gap was a tool that could run a site audit, score it, and produce guidance that non-technical stakeholders could understand and act on without a specialist interpreting the output.

How I worked

The technical stack: axe-core handles WCAG rule evaluation via Puppeteer, which crawls the target URL in a headless browser. Results are stored in Supabase and returned to the user through a Next.js frontend.

The AI layer takes the raw violation list and generates plain-English explanations of each issue, why it matters, and how to fix it. The output is designed to be handed directly to a developer with enough context to act without further clarification.

The PDF report was designed as a proper deliverable, not a raw data export. It includes a severity-weighted score, issue categories, and specific remediation guidance.

Results & impact

Working prototype complete. The tool runs live audits and produces reports that include severity scoring, categorised violations, and AI-generated fix guidance in plain English.

The project demonstrates product thinking applied to a real workflow problem: taking a manual, time-consuming process and turning it into something a client can run themselves in a few minutes.