ComfyUI

Provides local control to list nodes, read schemas, validate workflows, and execute ComfyUI workflows via MCP clients.
  • typescript

1

GitHub Stars

typescript

Language

5 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

This MCP server lets you integrate ComfyUI with AI assistants to list nodes, read node schemas, validate workflows, and execute them locally without starting an external HTTP server. It enables smooth, scriptable workflow orchestration between your ComfyUI installation and MCP clients like Claude.

How to use

You connect your MCP client to the ComfyUI MCP Server to access installed node types, fetch exact input/output schemas, and verify workflows before running them. Typical usage patterns include listing available nodes, validating a workflow JSON, and performing a dry run or a full run. Use dry runs to confirm steps without executing changes, then switch to a full run once you are confident.

To start using it, first ensure the MCP server is registered with your MCP client. The local server runs as a stdio process that you initiate from your environment, and it reads your ComfyUI installation location from an environment variable. Once connected, you can issue prompts like: list all nodes, validate this workflow JSON, or run this workflow with a specified number of steps.

How to install

Prerequisites you need before installing are straightforward.

  • Python 3.10+

  • Local ComfyUI installation

  • An MCP client such as Claude Code, Claude Desktop, or another compatible tool

Follow these steps to install the ComfyUI MCP Server locally.

git clone https://github.com/zturner1/comfyui-mcp.git
cd comfyui-mcp
pip install -e .

Configuration and usage with Claude clients

You configure the server so your MCP client can discover and communicate with ComfyUI. Set the path to your ComfyUI installation using an environment variable so the MCP server can locate the software.

# Windows (PowerShell)
$env:COMFYUI_ROOT = "D:\path\to\ComfyUI"

# Linux/macOS
export COMFYUI_ROOT="$HOME/ComfyUI"

Add ComfyUI as an MCP server in your client

Register the ComfyUI MCP server with your MCP client so you can start querying and running workflows.

claude mcp add comfyui -- comfyui-mcp

Example configuration for Claude Desktop

If you configure Claude Desktop manually, include the following MCP server entry in your client settings.

{
  "mcpServers": {
    "comfyui": {
      "command": "comfyui-mcp",
      "env": {
        "COMFYUI_ROOT": "/path/to/ComfyUI"
      }
    }
  }
}

Available tools

listNodes

Fetches all installed node types, including custom nodes, so you can see what is available for workflow construction.

readNodeSchema

Retrieves the exact input/output schemas for a specified node to ensure compatibility in your workflows.

validateWorkflow

Validates a workflow JSON before execution to catch schema or reference errors early.

executeWorkflow

Executes a validated workflow either as a dry run or a full run, with optional step reduction.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational
ComfyUI MCP Server - zturner1/comfyui-mcp | VeilStrat