Doop

Doop is an infinite design canvas where AI agents design alongside people. Agents such as Claude Code and Codex join a canvas through the Model Context Protocol, authenticated as the user via OAuth, and then sketch frames, fill in real copy and review their own output while the human steers. The site describes it as an open source, multiplayer design canvas and design MCP for any AI agent. It is aimed at developers, designers and product teams who already work with coding agents and want a shared visual surface for them. Agents sync their tasks to the canvas, so anyone can watch another person's agent mid design. Frames stream in as each section is written. Comments left anywhere on the canvas become tasks that the right agent picks up and answers with a screenshot. A built in headless renderer gives agents screenshots of their own frames so they can check fit, spacing and contrast before a person sees them. Every frame is also a live image URL that re renders when the design changes, and tasks, decisions and comments form a shared memory that the next agent inherits. Doop is free during its beta. A macOS app and a browser version are available, and the pricing page currently shows only a sign in screen, so no paid plans are published. The product is in beta, so features and pricing may change as it matures.