ZSeven-W/openpencil — reverse-engineered prompt

Build me an open source AI powered vector design app called OpenPencil that feels like a modern alternative to Pencil. I want a live infinite canvas where I can describe a UI in plain English and watch it appear in real time, then click existing elements and chat to revise them. It should support agent teams so bigger pages can be split into sections and generated at the same time, with clear visual feedback showing what each agent is working on. Please include style guides like glass, brutalist, and retro, support image or screenshot reference input, and save designs as readable .op JSON files that work well with Git.

I also want code export from one design into HTML and CSS, React, Vue, Svelte, Flutter, SwiftUI, Jetpack Compose, and React Native. Make it usable as a web app and a desktop app, with a simple CLI and MCP support so coding agents can read and edit .op files too. Look up current docs online if you need to.