Accessibility
Accessibility Linter Extension
An alt-text experiment became a broader page-level accessibility analysis workflow.
EdPlus at ASU
- Chrome Extension
- axe-core
- FastAPI
- Gemini
Measured
Two passes
Markup analysis and vision analysis reconciled against each other, rather than run in isolation.
Measured
Handed off
Code donated to another high impact project.
The Friction
A colleague was piloting an alt-text generator. It was useful, but alt text is one slice of accessibility, and the rest of the page was still a manual check. Worse, a generated alt text can be fluent and still describe the wrong image.
What I Did First
I joined the pilot to understand where it stopped and why. The gap was that markup analysis and visual analysis were being treated as separate problems when the interesting failures live between them.
What I Built
A Chrome extension that takes both an HTML snapshot and a screenshot of a page, runs a WCAG script against the markup and a vision pass against the image, then reconciles the two. That pairing catches mismatches neither finds alone, including alt text that does not describe the image it belongs to. Later additions included table heading detection, distinguishing decorative tables from real ones, and contrast ratio flagging.
What Came of It
The code was donated to another high impact project.
Scope and Limits
A proof of concept that never shipped. The handoff was the outcome. The extension surfaces issues and suggests fixes; remediation stayed out of scope on purpose, since a browser extension is the wrong place to rewrite someone's course.
More systems, and the learning experiences behind them.
or write directly to brent.michael670@gmail.com