How much of your Cursor usage is spent on tiny UI tweaks?

Discussions
1

I’ve noticed something while building with Cursor.

A surprisingly large amount of my prompts are not feature development. They’re things like:

  • make this button wider

  • move this card left

  • reduce this padding

  • change this text

  • make this section larger

Cursor does the edits correctly, but I still spend tokens and time describing visual changes that I can already see on the screen.

I’m curious:

How much of your Cursor usage goes into these kinds of UI iteration prompts versus actual feature development?

To explore this, I built a small open-source tool called Nuvio that lets me:

  • Click React components in the browser

  • Edit text and Tailwind styles visually

  • Preview the diff

  • Write the changes back to source code

Not posting this as a promotion — I’m genuinely trying to understand whether this is a problem other Cursor users have, or if it’s just my workflow.

Would love to hear how others handle UI iteration today.

5

Hi @Ehsan_ahmad Thanks for sharing!

Will be interesting to hear the community weigh in. IMHO, model selection is pretty important here. There are very few cases where you need to use frontier models for this type of work.

6

Hi @kevinn thank you for sharing your thoughts.

That’s a good point. I also find myself switching to cheaper models for a lot of UI work.

What got me thinking though was that even when the cost is low, I’m still spending time describing changes that are already visible on the screen. Things like spacing, alignment, sizing, and text updates feel more like direct manipulation problems than coding problems.

I’m curious whether others have developed workflows for this, or if most people are happy staying entirely inside the chat loop.