After back-and-forth chats with Cursor about frontend changes, I realized it's hard for me to describe a visual intent in text prompts. So I built a drop-in React component that lets you draw on your frontend & copy it as an image. Paste that in your coding agent for faster prompting :)