comfyui-mcp-server
4
ComfyUI MCP Server is designed to integrate ComfyUI with the Model Context Protocol, allowing for enhanced server functionality using built-in and custom workflows. It is a tool for managing and deploying workflows that convert text to images and more, deployed via UV or Docker.
ComfyUI MCP Server
- A server implementation for integrating ComfyUI with Model Context Protocol (MCP).
- Requires a running ComfyUI server; you need to host it or have access to one.
Features
- text_to_image: Returns the URL of the generated image. Use
download_image
to get the actual image. - download_image: Downloads images generated by other tools.
- run_workflow_with_file: Run a workflow by providing the path to a workflow JSON file.
- run_workflow_with_json: Run a workflow by providing the workflow JSON data directly.
Installation and Configuration
- Set ComfyUI host and port in
src/.env
. - Add custom workflows by placing JSON files in the
workflows
directory.
Running the Server
- Recommended to use UV for running the server.
- Use Docker for containerized deployments; special considerations needed for image data and URLs.
Debugging
- Debug ComfyUI with Python scripts.
- Debug MCP server using MCP development commands.