home / mcp / comfyui mcp server
Provides a local MCP server that generates and refines images, audio, and video by controlling a local ComfyUI instance through an API.
Configuration
View docs{
"mcpServers": {
"jaewilson07-comfy_mcp": {
"url": "http://127.0.0.1:9000/mcp"
}
}
}Create images from prompts using the MCP workflow with optional overrides for width, height, steps, model, and other generation parameters.
Produce audio content based on provided tags and lyrics, integrated into the MCP workflow.
Recreate an existing asset using stored workflow data, with optional parameter overrides.
Display generated images inline for quick inspection.
Check the current state of the ComfyUI queue to monitor running or pending jobs.
Poll the status of a specific job by prompt_id to determine completion.
Browse recently generated assets to enable memory and iteration across sessions.
Fetch full provenance and parameters for a particular asset, including workflow history.
Cancel a queued or running job to stop generation early.
List available ComfyUI models to help you choose the right model for your task.
Retrieve current default values used by the MCP server.
Set default values to avoid repeating common parameters, with optional persistence.
Enumerate all ComfyUI nodes with categories and metadata to aid workflow construction.
Provide detailed schema for a specific node, including inputs and outputs.
Search for nodes by name, category, or output type to assemble or extend workflows.
Obtain input parameter specifications for a node to troubleshoot configurations.
List all available workflows discovered in the workflows/ directory.
Execute any workflow with custom parameters to perform complex tasks.
Show publish status and environment readiness for asset distribution.
Define the output directory for ComfyUI assets to ensure consistent storage.
Publish a generated asset into the project’s web directory with deterministic naming and optional manifest integration.