Comfy MCP Server

🚀 Comfy MCP Server

A server that uses the FastMCP framework to generate images based on prompts through a remote Comfy server.

🚀 Quick Start

This script uses the FastMCP framework to set up a server that generates images based on prompts by interacting with a remote Comfy server through a specified workflow.

✨ Features

generate_image(prompt: str, ctx: Context) -> Image | str

This function generates an image based on the specified prompt. It follows these steps:

1. Check if all environment variables are set.
2. Load the prompt template from the JSON file.
3. Submit the prompt to the Comfy server.
4. Monitor the status of prompt processing.
5. Retrieve and return the generated image after the prompt is completed.

generate_prompt(topic: str, ctx: Context) -> str

This function generates a comprehensive image generation prompt based on the specified topic.

📦 Installation

Prerequisites

• uv package and Python project manager.
• A workflow file exported from Comfy UI. This code includes an example Flux-Dev-ComfyUI-Workflow.json, which is only used as a reference here. You need to export your own workflow and set the environment variables accordingly.

You can install the packages required for local development using the following command:

uvx mcp[cli]

📚 Documentation

Configuration

Set the following environment variables:

• Set COMFY_URL to point to your Comfy server URL.
• Set COMFY_WORKFLOW_JSON_FILE to the absolute path of the JSON file exported from the Comfy UI workflow.
• Set PROMPT_NODE_ID to the ID of the text prompt node.
• Set OUTPUT_NODE_ID to the ID of the output node that has the final image.
• Set OUTPUT_MODE to url or file to select the desired output.

Optional: If you run an Ollama server, you can connect to it to generate prompts.

• Set OLLAMA_API_BASE to the URL where Ollama is running.
• Set PROMPT_LLM to the name of the model hosted on Ollama for prompt generation.

Example:

export COMFY_URL=http://your-comfy-server-url:port
export COMFY_WORKFLOW_JSON_FILE=/path/to/the/comfyui_workflow_export.json
export PROMPT_NODE_ID=6 # Use the correct node ID
export OUTPUT_NODE_ID=9 # Use the correct node ID
export OUTPUT_MODE=file

Usage

Start the Comfy MCP server with the following command:

uvx comfy-mcp-server

Claude Desktop Configuration Example

{
  "mcpServers": {
    "Comfy MCP Server": {
      "command": "/path/to/uvx",
      "args": [
        "comfy-mcp-server"
      ],
      "env": {
        "COMFY_URL": "http://your-comfy-server-url:port",
        "COMFY_WORKFLOW_JSON_FILE": "/path/to/the/comfyui_workflow_export.json",
        "PROMPT_NODE_ID": "6",
        "OUTPUT_NODE_ID": "9",
        "OUTPUT_MODE": "file"
      }
    }
  }
}

🔧 Technical Details

Dependencies

• mcp: Used to set up the FastMCP server.
• json: Used to handle JSON data.
• urllib: Used to make HTTP requests.
• time: Used to add a delay when polling.
• os: Used to access environment variables.
• langchain: Used to create a simple LLM prompt chain to generate image generation prompts from topics.
• langchain-ollama: Used to integrate with Ollama.

📄 License

This project is protected by the MIT License.