# comfyui-mcp

> Local-first agent for ComfyUI generating images, video, audio via MCP server and Claude Code plugin. Works with Claude, ChatGPT, Gemini, or local LLMs.

- **Category:** AI
- **Licence:** MIT
- **Latest version:** v0.52.206
- **Open source:** oss
- **Languages:** Typescript
- **Tags:** Mcp, Mcp Server, Workflows, Offline, Local First, Image Generation, Ollama, Stable Diffusion, Comfyui, Self Hosted, Ai Agent, Model Context Protocol, Claude code, Natural language, Wan, Local llm, Agent skills, Video Generation, Flux, Comfyui Extension, Comfyui Mcp, Claude Plugin, Comfyui Mcp Server

## Description

comfyui-mcp is a local-first, agent-native control plane for ComfyUI that combines an MCP server with a Claude Code plugin to generate images, video and audio, author and run workflows, manage models and custom nodes through natural language interactions. It works with any LLM including Claude, ChatGPT, Gemini, or free local models via Ollama. The tools ship skills that provide model-specific expertise so agents know the right sampler, CFG, resolution, and model files for each architecture without trial and error. One config targets local installs, LAN, VPS, or Comfy Cloud. It runs on macOS, Linux, and Windows and auto-detects ComfyUI installations.

## Links

- [Homepage](https://github.com/artokun/comfyui-mcp)
- [Website](https://comfyui-mcp.artokun.io/docs)
- [GitHub](https://github.com/artokun/comfyui-mcp)
- [Discord](https://discord.gg/cW9arBhzCu)

---
Source: https://selfhost.directory/project/comfyui-mcp