MCP.so
ログイン
C

Comfyui Mcp

@artokun

Comfyui Mcp について

MCP server + Claude Code plugin for ComfyUI — execute workflows, generate images, visualize pipelines as Mermaid diagrams, compose/validate workflows, manage and download models, control VRAM, and explore custom nodes. 36 MCP tools. Cross-platform (macOS/Linux/Windows) with auto-

基本情報

カテゴリ

開発者ツール

トランスポート

stdio

公開者

artokun

投稿者

Arthur R Longbottom

設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "comfyui": {
      "command": "npx",
      "args": [
        "-y",
        "comfyui-mcp"
      ]
    }
  }
}

ツール

ツールは検出されませんでした

ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。

概要

What is Comfyui Mcp?

Comfyui Mcp is an MCP server and Claude Code plugin for ComfyUI that lets you execute workflows, generate images, visualize pipelines, manage models, control VRAM, and explore custom nodes directly from an AI coding assistant. It auto‑detects your ComfyUI installation and port, works on macOS, Linux, and Windows, and requires Node.js ≥ 22.0.0.

How to use Comfyui Mcp?

Add the MCP server to your Claude Code config (~/.claude/settings.json) with the command npx -y comfyui-mcp (optional CIVITAI_API_TOKEN environment variable). With ComfyUI running, ask Claude to generate an image – it will find or download a checkpoint, build a workflow, execute it, and return the image. You can also install the Claude Code plugin with claude plugin install comfyui-mcp to access slash commands, skills, agents, and hooks.

Key features of Comfyui Mcp

  • 80+ MCP tools covering generation, workflow execution, model management, and diagnostics
  • 10 slash commands: /comfy:gen, /comfy:viz, /comfy:debug, /comfy:batch, and more
  • 6 built‑in knowledge skills for prompt engineering, troubleshooting, model compatibility
  • 3 autonomous agents: explorer, debugger, and optimizer
  • 3 hooks including VRAM watchdog and job completion notifications
  • Local generation tracker with SQLite database for statistics and settings suggestions

Use cases of Comfyui Mcp

  • Generate images from natural language prompts with automatic workflow building
  • Visualize and debug ComfyUI workflows directly in the chat interface
  • Manage models: search, download, and list checkpoints, LoRAs, VAEs, etc.
  • Perform batch parameter sweeps across cfg, sampler, steps, and seed
  • Diagnose workflow failures with autonomous debugger agent

FAQ from Comfyui Mcp

What scope does Comfyui Mcp support?

It supports local and remote ComfyUI installations (Mac/Linux/Windows, RunPod, VPS, LAN). For Comfy Cloud, a community cloud‑mode is available by setting COMFYUI_API_KEY, but the primary target is self‑hosted instances.

What are the runtime requirements?

Node.js ≥ 22.0.0 and a running ComfyUI instance (local or remote). The server auto‑detects the ComfyUI port if installed locally.

Where does generation history data live?

Every enqueue_workflow call logs settings to a local SQLite database (generations.db). You can view stats with npm run generations:stats.

How do I manage VRAM and ComfyUI processes?

Tools include clear_vram (calls ComfyUI's /free endpoint), stop_comfyui, start_comfyui, and restart_comfyui – all with hooks that warn before stopping if there are unsaved changes.

What transport and authentication does Comfyui Mcp use?

It runs as a standalone MCP server over stdio (launched via npx). No built‑in authentication is required; an optional CIVITAI_API_TOKEN can be set for model search/download from Civitai.

コメント

「開発者ツール」の他のコンテンツ