Comfy-Org/comfy-mcp

Local MCP server for ComfyUI — run your local ComfyUI from AI agents

What it solves

Comfy MCP allows AI agents—such as Claude Code, Claude Desktop, or Cursor—to directly control your local ComfyUI installation. It bridges the gap between your AI assistant and your image generation workflow, enabling the agent to trigger generations, monitor progress, and manage your ComfyUI environment without you needing to manually switch between windows.

How it works

This project acts as an MCP (Model Context Protocol) server that runs as a subprocess on your machine. It wraps the comfy-cli engine, translating natural language requests from your AI agent into specific commands for your local ComfyUI instance. When you ask your agent to generate an image or check your system status, the server executes the corresponding comfy command, parses the output, and returns the results to your agent.

Who it’s for

It is designed for users who already run ComfyUI locally and want to integrate their image generation workflows into AI-powered coding or chat environments. It is ideal for those who want to automate tasks like running workflows, searching for models, or managing their ComfyUI server state through an AI interface.

Highlights

  • Direct Control: Run workflows, monitor job progress, and cancel tasks directly from your AI client.
  • System Introspection: Search your installed nodes, models, and templates to see exactly what your local setup can do.
  • Workflow Management: Validate graphs, edit template slots, and automate the creation of image variants.
  • Environment Management: Start, stop, and restart your ComfyUI server, or tail logs without leaving your AI chat.
  • Flexible Execution: While local-first, it supports running partner models on external infrastructure and can connect to remote ComfyUI instances you control.

Related

  • Project
    ATH-MaaS/Ovis-Image

    Ovis-Image is a 7B text-to-image model optimized for high-quality text rendering and layout precision, designed to run efficiently on accessible hardware.

  • Project
    Aixio-AI/aixio-layer-v2

    Aixio Layer v2 is an AI model that converts flat images into fully editable PowerPoint files, allowing users to manipulate individual text, shapes, and image layers.

  • Project
    yejy53/RealGen

    RealGen is a text-to-image generator that uses AI detectors as reward signals during reinforcement learning to produce highly photorealistic images with fewer artifacts.

  • Project
    imglib/imglib2

    ImgLib2 is a general-purpose, multidimensional image processing library for Java. It provides an interface-driven design that works with many data types and storage layouts, enabling dimension-independent algorithms with transparent data access, so developers can focus on the logic rather than low-level details.