comfyui-mcp is a local-first, agent-native control plane for ComfyUI, enabling AI agents to generate media, author workflows, and manage models.
Why it matters
overview
comfyui-mcp is an AI control plane tool developed by Art Longbottom (Maintainer) that enables developers and users of ComfyUI to drive their live graph in natural language using AI agents. It functions as an MCP server combined with an autonomous sidebar agent, facilitating the generation of images, video, audio, and 3D content, as well as the authoring and management of ComfyUI workflows and models.
features
comfyui-mcp provides a comprehensive set of features designed to streamline interaction with ComfyUI through natural language and AI agents. Its architecture includes an MCP server and an autonomous sidebar agent, facilitating advanced control and content generation.
use cases
comfyui-mcp is primarily designed for developers and users who seek to leverage AI agents for controlling ComfyUI, simplifying complex generative AI tasks, and automating content creation workflows.
how to use
To begin using comfyui-mcp, users typically install the MCP server and integrate it with their preferred LLM, which then drives the ComfyUI graph through natural language commands. The system translates these commands into executable ComfyUI operations.
artokun/comfyui-mcp server on a macOS, Linux, or Windows system.pricing
The artokun/comfyui-mcp project operates on a freemium model, with its core components being open-source. While the artokun/comfyui-mcp server itself is open-source and free to use, its functionality often relies on integration with external LLMs, which may incur costs. For instance, using cloud-based LLMs like Claude, ChatGPT, or Gemini typically requires a user's existing subscription or API key, which may involve usage-based fees. The official Comfy Cloud MCP, a related but distinct offering, operates on a credit-based system, with 10,000 credits equivalent to $1. New accounts receive 5,000 free credits. Paid plans include Basic ($10/month for 100k credits), Standard ($30/month for 315k credits), and Pro ($60/month for 660k credits), offering varying GPU access and priority.
Similar Tools
comfyui-mcp distinguishes itself in the generative AI landscape by providing a local-first, agent-native control plane for ComfyUI, emphasizing natural language interaction and broad LLM compatibility. It aims to simplify the complex node-based interface of ComfyUI for a wider audience.
It is an open-source AI assistant that simplifies ComfyUI workflow development by using natural language for node suggestions, workflow building, and parameter exploration, integrating directly into the UI.
Similar to comfyui-mcp, ComfyUI-Copilot offers natural language control for building and exploring ComfyUI workflows. It functions as an integrated UI assistant, whereas comfyui-mcp is described as a 'control plane' with an autonomous sidebar agent.
This custom node generates ComfyUI workflows directly from natural language descriptions by leveraging Large Language Models (LLMs), including local GGUF models.
ComfyUI-WorkflowGenerator directly competes by translating natural language into ComfyUI workflows, a core function of comfyui-mcp. As a custom node, it is local-first and free, aligning with comfyui-mcp's freemium and local-first approach, but operates as a node within ComfyUI rather than an external agent.
It provides a comprehensive LLM agent framework for ComfyUI, including an MCP server, that supports a wide array of local and cloud LLMs for workflow construction and integration.
ComfyUI LLM Party is highly aligned with comfyui-mcp, offering an MCP server and broad LLM compatibility (including Ollama for local models) to build and integrate LLM workflows within ComfyUI. It emphasizes building AI assistants and complex agent interactions, directly competing with comfyui-mcp's agent-native control plane.
ComfyAgent is an LLM-based agent framework designed to autonomously generate and design collaborative AI systems by creating ComfyUI workflows from task instructions.
This framework directly addresses the core problem of LLM-driven ComfyUI workflow generation, similar to comfyui-mcp's agent-native control plane. However, ComfyAgent is presented more as a research framework for designing AI systems rather than a direct user-facing product with a sidebar agent.
More on Stork
Other tools in this category, matched by shared tags