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 is a multiplayer design canvas that allows humans and AI agents to collaborate on UI/UX designs in real-time. It replaces the traditional "prompt-and-refresh" workflow by allowing agents to stream their design changes directly onto a shared canvas, where humans can watch and edit alongside them.

How it works

The platform uses a shareable canvas containing "Frames" (artboards that render real HTML in sandboxed iframes). It provides a built-in MCP (Model Context Protocol) server, allowing external AI clients like Claude Code to connect and edit the canvas live. It also features a "Doop Agent"—a server-side AI team with specialized roles (UX, copy, brand, accessibility) that can be triggered via chat mentions or task cards.

Who it’s for

Designers, developers, and product teams who want to integrate AI agents into their design process as active collaborators rather than just prompt-based generators.

Highlights

  • Real-time Collaboration: Live cursors, presence indicators, and an activity feed for both humans and agents.
  • MCP Integration: Connects to MCP clients (e.g., Claude Code) to let agents work as the user.
  • Built-in Design Team: Specialized server-side agents that can be routed through a pipeline of roles.
  • Design Memory: A "guideline distiller" that proposes durable style rules for agents to follow based on exemplar frames.
  • Flexible Deployment: Self-hostable via Docker or Bun with an embedded Postgres (PGlite) database.
  • External Integrations: Supports importing from websites, live apps, and GitHub repos, and integrates with Linear for ticket-based design tasks.

Related

  • Project
    Minidoracat/mcp-feedback-enhanced

    An MCP server that inserts human checkpoints into AI workflows, allowing users to provide real-time feedback via a Web UI or desktop app to reduce AI drift.

  • Project
    docker/compose-for-agents

    A collection of Docker Compose demos for deploying and orchestrating AI agents, supporting multiple frameworks and local or remote model execution.

  • Project
    lgazo/drawio-mcp-server

    An MCP server that enables AI agents to programmatically create and modify Draw.io diagrams, automating the creation of architectural diagrams and flowcharts.

  • Project
    agentrq/agentrq

    An agent-human collaboration platform that uses the Model Context Protocol (MCP) to synchronize AI agent tasks with a visual workspace and remote machine execution.

  • Project
    Team-Commonly/commonly

    An open-source coordination workspace where humans and AI agents collaborate using shared pods, persistent memory, and a GitHub-synced task board.