ComfyUI Agent Control
๐ ๐ - MCP server for ComfyUI that lets AI agents view, edit, and run node-based image generation workflows with an embedded terminal.
What is ComfyUI Agent Control?
ComfyUI Agent Control is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to ๐ ๐ - mcp server for comfyui that lets ai agents view, edit, and run node-based image generation workflows with an embedded terminal.
๐ ๐ - MCP server for ComfyUI that lets AI agents view, edit, and run node-based image generation workflows with an embedded terminal.
This server falls under the Developer Tools category on MCPgee, the world's largest MCP server directory with 33,000+ servers.
Features
- ๐ ๐ - MCP server for ComfyUI that lets AI agents view, edit,
Use Cases
Maintainer
Works with
Installation
Manual Installation
npx comfy-pilotConfiguration
Configuration Details
claude_desktop_config.json
Performance
Response Metrics
Resource Usage
How to Set Up and Use ComfyUI Agent Control
Comfy Pilot is an MCP server implemented as a ComfyUI custom node that gives AI agents direct control over ComfyUI's node-based image generation environment. It exposes 13 tools covering workflow inspection, node editing, image preview, custom-node management, and model downloads, so an agent like Claude can read, modify, and execute Stable Diffusion workflows without a human touching the UI. Developers and AI artists who want to drive ComfyUI programmatically โ including automating iterative image generation pipelines โ will find this the most direct integration path.
Prerequisites
- Python 3.10 or newer with ComfyUI installed and running locally
- ComfyUI Manager installed (recommended for easiest node installation)
- Claude Desktop or another MCP-capable client
- Git available if cloning manually
- Optional: CivitAI or Hugging Face account for model downloads via the server
Install Comfy Pilot as a ComfyUI custom node
Use the ComfyUI CLI to install the Comfy Pilot node directly into your ComfyUI custom_nodes directory. Alternatively, install via ComfyUI Manager by searching 'comfy-pilot' in the UI, or clone the repository manually.
comfy node install comfy-pilotLocate the MCP server script
After installation, the MCP server entry point is at mcp_server.py inside the comfy-pilot directory under ComfyUI's custom_nodes folder. Note the full absolute path โ you will need it for the client configuration.
ls ~/.local/share/ComfyUI/custom_nodes/comfy-pilot/mcp_server.pyStart ComfyUI
Make sure ComfyUI is running before connecting the MCP client, as the server communicates with the live ComfyUI instance over its local API.
python main.py --listen 127.0.0.1 --port 8188Configure your MCP client
Add the Comfy Pilot server to your Claude Desktop or other MCP client configuration, pointing to the mcp_server.py script. Replace the path with your actual installation location.
{
"mcpServers": {
"comfyui": {
"command": "python3",
"args": ["/path/to/ComfyUI/custom_nodes/comfy-pilot/mcp_server.py"]
}
}
}Restart your MCP client
Restart Claude Desktop (or your chosen MCP client) so it picks up the new server entry. The comfyui server should appear in the connected tools list.
Verify the connection
Ask the AI to list the nodes in the current ComfyUI workflow. If the server is connected correctly, it will return the node graph data from the running ComfyUI instance.
ComfyUI Agent Control Examples
Client configuration
Minimal claude_desktop_config.json entry for Comfy Pilot. Adjust the path to match your ComfyUI installation.
{
"mcpServers": {
"comfyui": {
"command": "python3",
"args": ["/home/user/ComfyUI/custom_nodes/comfy-pilot/mcp_server.py"]
}
}
}Prompts to try
These prompts exercise the 13 tools exposed by Comfy Pilot, from reading the current workflow to downloading models.
- "What nodes are in my current ComfyUI workflow and how are they connected?"
- "Add a KSampler node and connect it to the existing checkpoint loader"
- "Run the workflow and show me the output image"
- "Download the FLUX.1 schnell model from Hugging Face"
- "Search for available ControlNet custom nodes and install the best one"
- "Run the workflow only up to node 5 and show me the intermediate result"Troubleshooting ComfyUI Agent Control
Server starts but returns errors when calling tools
Ensure ComfyUI is running on http://127.0.0.1:8188 before the MCP client connects. Comfy Pilot communicates with ComfyUI's HTTP API, so the ComfyUI process must be active.
python3 not found or wrong Python version used
Specify the full path to the Python interpreter that has ComfyUI's dependencies installed (e.g., /home/user/ComfyUI/venv/bin/python3) in the args of the MCP config instead of a bare python3.
Custom node not visible in ComfyUI after installation
Restart ComfyUI after running 'comfy node install comfy-pilot'. ComfyUI loads custom nodes only at startup. If the node still doesn't appear, confirm the directory was created under custom_nodes and there are no import errors in the ComfyUI console.
Frequently Asked Questions about ComfyUI Agent Control
What is ComfyUI Agent Control?
ComfyUI Agent Control is a Model Context Protocol (MCP) server that ๐ ๐ - mcp server for comfyui that lets ai agents view, edit, and run node-based image generation workflows with an embedded terminal. It connects AI assistants to external tools and data sources through a standardized interface.
How do I install ComfyUI Agent Control?
Follow the installation instructions on the ComfyUI Agent Control GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.
Which AI clients work with ComfyUI Agent Control?
ComfyUI Agent Control works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.
Is ComfyUI Agent Control free to use?
Yes, ComfyUI Agent Control is open source and available under the MIT license. You can use it freely in both personal and commercial projects.
ComfyUI Agent Control Alternatives โ Similar Developer Tools Servers
Looking for alternatives to ComfyUI Agent Control? Here are other popular developer tools servers you can use with Claude, Cursor, and VS Code.
Ecc
โ 188.2kThe agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
Javaguide
โ 155.8kJava ้ข่ฏ & ๅ็ซฏ้็จ้ข่ฏๆๅ๏ผ่ฆ็่ฎก็ฎๆบๅบ็กใๆฐๆฎๅบใๅๅธๅผใ้ซๅนถๅใ็ณป็ป่ฎพ่ฎกไธ AI ๅบ็จๅผๅ
Gemini CLI
โ 104.5kA secure MCP server that wraps the Google Gemini CLI, allowing clients to query Gemini models using local OAuth sessions without requiring an API key. It provides tools for model interaction and diagnostics with built-in protection against command in
Awesome MCP Servers
โ 87.3kโญ Curated list of Model Context Protocol (MCP) servers - tools that extend Claude Desktop, Cursor, Windsurf, and other MCP clients with custom capabilities.
MCP Servers
โ 86.0kModel Context Protocol Servers
CC Switch
โ 77.5kA cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Gemini CLI & Hermes Agent. Only official website: ccswitch.io
Browse More Developer Tools MCP Servers
Explore all developer tools servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.
Set Up ComfyUI Agent Control in Your Editor
Choose your AI client for step-by-step setup instructions.
Quick Config Preview
Add this to your claude_desktop_config.json or .cursor/mcp.json
Ready to use ComfyUI Agent Control?
Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.