ComfyUI

You deploy this DSL-first MCP server to let AI agents manage ComfyUI workflows using a readable DSL. It converts between DSL and ComfyUI JSON automatically, enabling streamlined workflow creation, execution, and inspection from your agent workflows.

How to use

You interact with the MCP server through an MCP client that speaks the Model Context Protocol. Define or load DSL workflows, convert them to ComfyUI JSON, and execute them from your AI agent conversations. You can read existing workflows, validate DSL syntax, and monitor job status as workflows run on ComfyUI. Use the client to ask for the current queue, to start a workflow, or to inspect generated outputs.

How to install

Prerequisites: Python is required to install and run the MCP server.

Install the MCP package with Python’s package manager.

pip install comfy-mcp

Configuration and operation

Configure the MCP server you will run locally. The following is the explicit MCP server configuration you can use to start managing ComfyUI workflows.

{
  "mcpServers": {
    "comfyui-workflows": {
      "command": "comfy-mcp",
      "args": [],
      "env": {}
    }
  }
}

Available tools

read_workflow

Read and convert workflows to DSL (reads a workflow JSON and returns its DSL)

write_workflow

Write DSL to disk as JSON/DSL (saves the current DSL to a file)

list_workflows

Discover workflow files in a directory or repository

validate_workflow

Check DSL syntax for correctness and consistency

get_workflow_info

Analyze DSL structure to extract metadata about a workflow

execute_workflow

Run a DSL workflow on the ComfyUI server

get_job_status

Monitor execution and retrieve status or outputs (images)

list_comfyui_queue

View current ComfyUI queue status