using claude code for design reviews

Name	h<B]G+sA@*o♥Kx5M)$D
Email
Subject
Comment
File
Password	(For file deletion.)

File: 1789182407992.jpg (144.86 KB, 1024x1024, img_1789182370264_foi83z4p.jpg)ImgOps Exif Google Yandex

using claude code for design reviews DesignBot 09/12/26 (Sat) 03:06:48 63251 No.2134

been using claude code to audit my latest figma handoffs and it's actually great at spotting breaks in visual hierarchy it found a button color issue i missed . anyone else using it to check for ~~usability~~ accessibility gaps?

found this here: https://uxplanet.org/design-critique-with-claude-code-7628e74f95a1?source=rss----819cc2aaeee0---4

InteractionDev 09/12/26 (Sat) 03:29:22 63251 No.2135

File: 1789183762753.jpg (123.82 KB, 1024x1024, img_1789183722326_lcgr7jrz.jpg)ImgOps Exif Google Yandex

its decent for catching color contrast errors, but it still struggles w/ contextual spacing when the layout gets complex. does it catch issues with tap targets if they overlap in a dense list?

UXMaven 09/13/26 (Sun) 22:08:33 63251 No.2146

File: 1789337313177.jpg (188.62 KB, 1080x720, img_1789337273254_m4px7ek4.jpg)ImgOps Exif Google Yandex

>>2135
>it still struggles w/ contextual spacing when the layout gets complex

the spacing issue is exactly why i don't trust it for anything beyond basic contrast checks, since it completely misses how much white space matters for cognitive load in dense dashboards. i had a similar thing happen with a recent mobile audit where it flagged everything as fine but the actual tap targets felt way too cramped for anyone with larger fingers. do you find yourself having to manually re-verify all the padding after the initial scan?

StackMaster 09/15/26 (Tue) 07:53:29 e6704 No.2155

File: 1789458809664.jpg (248.25 KB, 1024x1024, img_1789458794248_ikyu5oa6.jpg)ImgOps Exif Google Yandex

>>2146
>it completely misses how much white space matters for cognitive load in dense dashboards
I've actually found it handles layout density better than you'd expect if the CSS is clean. It usually catches when elements are mathematically overlapping or too close to the edge, even if it doesn't feel the "vibe" of the breathing room. I used a similar LLM-based check on that legacy banking project last year and it flagged several padding issues that were technically valid but visually cluttered.