Frontend

woah. I just learnt about accessibility in DOM nodes. this is very neat. Inspect Elements -> 🚹 icon top right in the panel.

The WAI-ARIA accessibility tree is underrated, and it’s why web browser engines are still the best form factor for any serious multi-modal software application IMO.

Also why it’s absolutely crucial to rely on semantic HTML elements and built-in browser behaviors, instead of replicating everything with `<div>`s and JavaScript and smoke and mirrors (which many site-builders and vibe-coding tools are prone to perpetuate).

The WAI-ARIA accessibility tree is underrated, and it’s why web browser engines are still the best form factor for any serious multi-modal software application IMO.

Also why it’s absolutely crucial to rely on semantic HTML elements and built-in browser behaviors, instead of replicating everything with `<div>`s and JavaScript and smoke and mirrors (which many site-builders and vibe-coding tools are prone to perpetuate).

https://web.dev/articles/the-accessibility-tree

https://developer.chrome.com/blog/full-accessibility-tree/

https://developer.mozilla.org/en-US/docs/Glossary/Accessibility_tree

/frontend dev (HTML/CSS)・founder/decentralized data wrangler /blockhead・contributor /ethglobal /walletbeat・hackathon bounty hunter・music maker・dropout・nomad・25

lol can these agents just run through your codebase in the CI and update the semantic markup parsing through this accessibility DOM? I'm sure alums can figure out what's a header, what's body copy, what's an accent?

software artist | building /outpaint | ex: @coinbase ex: bisontrails | seeker of the flow state | artivilla.com

If you aren’t adhering to semantics to begin with, the accessibility tree is kinda useless

Maybe computer vision models and clever CSS class analysis can bridge that gap, but ideally you and your LLMs should be building with semantic building blocks from the start and not as an afterthought