Comfyui Mcp

Please install the `comfyui-mcp` MCP server into my current AI client (that's you).

Required prerequisites (do these first if not already done):
- **Install Node.js** — Node.js 22.0.0 or later is required to run the MCP server via npx. (https://nodejs.org/)
- **Install and run ComfyUI** — Install ComfyUI and have it running before using the server. Local mode is the default; remote or cloud modes are also supported via environment variables. (https://github.com/comfyanonymous/ComfyUI)

Canonical MCP server config (stdio transport):
- command: `npx`
- args: ["-y","comfyui-mcp"]
- optional environment variables:
  - `CIVITAI_API_TOKEN`: CivitAI API token for model downloads. (example: `<your-civitai-api-token>`)
  - `COMFYUI_URL`: Full ComfyUI URL to target a specific local or remote ComfyUI instance. Overrides COMFYUI_HOST/COMFYUI_PORT and skips auto-detection. (example: `<your-comfyui-url>`)
  - `COMFYUI_HOST`: ComfyUI server host when not using COMFYUI_URL. (example: `<your-comfyui-host>`)
  - `COMFYUI_PORT`: ComfyUI server port when not using COMFYUI_URL. (example: `<your-comfyui-port>`)
  - `COMFYUI_PATH`: Path to the ComfyUI data directory containing models and related files. Useful when auto-detection fails. (example: `<your-comfyui-path>`)
  - `COMFYUI_API_KEY`: Comfy Cloud API key. When set, cloud mode is activated and the server talks to Comfy Cloud instead of a local/remote ComfyUI instance. (example: `<your-comfyui-api-key>`)
  - `COMFYUI_CLOUD_URL`: Override the Comfy Cloud endpoint for testing or staging. (example: `<your-comfyui-cloud-url>`)
  - `HUGGINGFACE_TOKEN`: HuggingFace token for higher API rate limits or gated model access. (example: `<your-huggingface-token>`)
  - `GITHUB_TOKEN`: GitHub token used for skill generation to avoid rate limits. (example: `<your-github-token>`)
  - `REGISTRY_ACCESS_TOKEN`: Comfy Registry API key for publish_custom_node functionality. (example: `<your-registry-access-token>`)
  - `COMFYUI_DOWNLOAD_CACHE_DIR`: Directory for the content-addressed model download cache. (example: `<your-download-cache-dir>`)
  - `COMFYUI_LRU_CACHE_SIZE_GB`: Maximum download cache size in GB. Set to limit cache growth; 0 disables LRU eviction. (example: `<your-cache-size-gb>`)
  - `COMFYUI_STARTUP_CHECK_INTERVAL_S`: Readiness probe interval in seconds when starting a local ComfyUI process. (example: `<your-startup-check-interval-seconds>`)
  - `COMFYUI_STARTUP_CHECK_MAX_TRIES`: Maximum readiness probe attempts when starting a local ComfyUI process. (example: `<your-startup-check-max-tries>`)
  - `COMFYUI_ALWAYS_RESTART`: Whether to auto-restart a crashed local ComfyUI process. (example: `<true-or-false>`)
  - `LOG_LEVEL`: Logging verbosity. (example: `<debug-or-info-or-warn-or-error>`)

Note: Default transport is stdio via `npx -y comfyui-mcp`. The server auto-detects ComfyUI installation path and port. It also supports HTTP mode with `--http` or `MCP_TRANSPORT=http`, but the simplest documented canonical config is stdio. In cloud mode, set `COMFYUI_API_KEY`. In remote mode, set `COMFYUI_URL` or pass `--comfyui-url`.

Add this MCP server to my current client's config in the correct format for you. If you need secrets or credentials I haven't provided, ASK me — do not invent values or leave raw placeholders. After adding it, tell me how to verify the server is connected.