Post

Responsive design is much easier with AI. But I still have to tell my agent *how* to be responsive; what content is overflowing, which elements should wrap in which way, when a table should be horizontally scrollable etc. I cannot just say "make this responsive", I still need to provide my taste.

The agent copied the breakpoints the rest of the repo uses. In a modal the viewport is the wrong axis by construction, and asking for the right breakpoint in the prompt only fixes that one component. Making container queries the default in the repo is the part that holds.