Is a Cursor project a valid design handoff for a developer using Claude AI

I’m a UI/UX designer working on a major redesign of an existing app that has a frontend and backend already in production. Almost every screen looks different but the functionality stays the same — nothing was added or removed, only the UI was redesigned from scratch. A few button labels and action names were also renamed for better UX clarity — same actions, clearer wording.

I do my final designs in Cursor, not Figma. I have components, styles, spacing, colors, and interactions all built there.

The frontend developer uses Claude AI to implement changes into the existing codebase. I was told that the developer feeds Claude screenshots and prompts to implement the changes — no Figma links, only screenshots and prompts.

I was specifically asked to refine the app and make it more user friendly and deliver screenshots. Instead of spending time doing frame by frame in Figma, I made 5-8 frames in Figma on mobile and desktop to define how I wanted the elements to be positioned and structured. I then built it fully in Cursor with components and styles to have a consistent design system and a clear handoff for 50+ screens across mobile and desktop.

I then prepared before/after screenshots of all screens organized in clear flow paths covering all scenarios — more than 50 screenshots covering both mobile and desktop, including screenshots taken directly from the current production version alongside my redesigned versions.

My work was dismissed with the explanation that it cannot be implemented in the existing frontend and backend. But I was specifically asked to refine and improve the UI — not rebuild functionality. The design is better, the functions are identical.

Given that the developer uses only screenshots and prompts fed directly to Claude — no Figma links — is there any technical reason why screenshots exported from a Cursor project would be less useful than screenshots from Figma?

Is the before/after screenshot approach a valid handoff for Claude-assisted frontend implementation? And what exactly should I deliver to make the developer’s job as easy as possible?

Maybe he doesn’t want to pay for Figma’s paid plan, which allows him to connect Claude to Figma for direct communication between the two systems.

You’ve already asked GPT-5.4 in Cursor to explain “why” to you, haven’t you?

To add more context: I was specifically asked to deliver in Figma, and I did. I delivered:

  • Low-fidelity frames in Figma showing element positioning and structure

  • Before/after screenshots of all 50+ screens, mobile and desktop, including screenshots from the current production version

  • The final high-fidelity design built in Cursor with full components, styles, interactions and animations

The Figma paid plan isn’t the issue—I was already working in Figma as requested. The Cursor build was an addition to make the handoff more complete, not a replacement for Figma.

Given all of this — before/after screenshots, low fidelity Figma frames, and a fully built Cursor prototype — what exactly would still be missing for a developer using Claude AI to implement the changes?

My conversations were actually with Claude :shushing_face: , trying to understand the handoff problem myself — we both got confused, which is why I posted here. Thanks for reaching out, by the way.:heart: