kgoedecke/doop

The open-source alternative to Paper.design. A multiplayer design canvas where humans and AI agents design together, live. MCP built in.

What it solves

Doop provides a multiplayer design canvas where humans and AI agents can collaborate in real-time. It eliminates the "prompt-and-refresh" cycle by allowing agents to stream their designs directly onto a shared canvas, enabling users to watch AI work and provide immediate feedback.

How it works

The platform uses a shared canvas containing "Frames" (sandboxed iframes rendering real HTML). Humans edit via a browser interface, while AI agents connect through a built-in Model Context Protocol (MCP) server. The server manages live synchronization via WebSockets, handling cursors, presence, and a "typewriter reveal" that smooths out agent-generated HTML streams into a steady visual flow for human viewers.

Who it’s for

  • UI/UX Designers who want to collaborate with AI agents on web layouts.
  • Developers looking for an open-source alternative to collaborative design tools like Paper.design.
  • AI Agent developers wanting to integrate LLMs into a visual, multiplayer workspace.

Highlights

  • MCP Integration: Connects seamlessly with Claude Code or any MCP client to allow agents to act as first-class collaborators.
  • Llive Streaming Designs: Agents stream HTML in chunks, which the server renders as a smooth live reveal with a pulsing border to indicate active work.
  • Multiplayer Experience: Features live cursors, presence indicators, and a shared activity feed for all human and agent actions.
  • Agent Steering: Users can leave feedback on specific agent tasks, which the next agent to interact with the canvas claims and addresses.
  • Self-Hostable: Can be deployed quickly via Docker or npm with an embedded Postgres (PGlite) database for zero-config startup.

Related

  • Project
    different-ai/openwork

    An open-source desktop app and MCP server for sharing and reusing AI workflows, skills, and connected services across different AI agents and teams.

  • Project
    21st-dev/magic-mcp

    An MCP server that connects AI agents to the 21st.dev platform for searching, generating, and managing UI components and templates.

  • Project
    kanwas-ai/kanwas

    A multiplayer workspace for teams and AI agents to collaborate on shared documents and evidence using a Git-backed markdown filesystem.

  • Project
    xintaofei/codeg

    A multi-agent coding workspace that aggregates various AI coding agents into a single interface, enabling collaboration and collaboration and task management via git worktrees.

  • Project
    hi-godot/godot-ai

    A plugin that connects AI assistants to the Godot editor via the Model Context Protocol (MCP), allowing AI agents to programmatically build scenes, edit scripts, and configure game environments.