广告合作
MCP.so
登录

ComfyUI MCP Server

@Overseer66

代码仓库

基本信息

分类

其他

许可证

Apache-2.0

传输方式

stdio

作者

Overseer66

关于 ComfyUI MCP Server

comfyui-mcp-server

ComfyUI MCP Server 配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "comfyui-mcp-server-overseer66": {
      "command": "python",
      "args": [
        "src/test_comfyui.py"
      ]
    }
  }
}

ComfyUI MCP Server 工具

工具列表暂未提供

当前收录暂时没有可展示的工具列表，不代表服务没有工具。请查阅项目文档了解支持的能力。

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is ComfyUI MCP Server?

A server implementation for integrating ComfyUI with the Model Context Protocol (MCP). It requires a running ComfyUI server—either self-hosted or an existing server address—and provides tools to generate images and run workflows via MCP.

How to use ComfyUI MCP Server?

Configure the ComfyUI host and port in src/.env, place custom workflow JSON files in the workflows directory, then run the server using UV (with mcp run src/server.py:mcp) or Docker (using the prebuilt image overseer66/mcp-comfyui). Debug locally with python src/test_comfyui.py or mcp dev src/server.py.

Key features of ComfyUI MCP Server

  • Integrates ComfyUI with MCP for AI agent workflows.
  • Built-in tools: text_to_image, download_image, run_workflow_with_file, run_workflow_with_json.
  • Custom workflows added by placing JSON files in the workflows directory.
  • Supports UV and Docker deployment.
  • Offers SSE transport for remote connections.
  • Debugging scripts for ComfyUI and MCP are included.

Use cases of ComfyUI MCP Server

  • Generate images by running ComfyUI text-to-image workflows from an MCP client.
  • Download generated images via the download_image tool after a text_to_image call.
  • Execute arbitrary ComfyUI workflows by providing a JSON file path or inline JSON data.
  • Run ComfyUI workflows inside a Docker container for isolated or remote environments.
  • Connect to a remote ComfyUI server via SSE transport for cloud-based setups.

常见问题

1

What dependencies are required?

A running ComfyUI server (self-hosted or accessible) and Python packages: mcp, websocket-client, and python-dotenv.

2

How do I add a new workflow as a tool?

Place the workflow JSON file in the `workflows` directory and declare it as a new tool in the system.

3

Can I use Docker? Are there limitations?

Yes. When using Docker, the host filesystem is not shared, so downloading images to a local folder may be difficult. Set `RETURN_URL=false` to receive image data as bytes, and set `COMFYUI_HOST` to an appropriate address (e.g., `host.docker.internal`). Note that large image payloads may exceed response limits.

4

How do I connect via SSE transport?

Run the SSE server with Docker (`docker run -i --rm -p 8001:8000 overseer66/mcp-comfyui-sse`) and set the `url` in `mcp.json` to `http://localhost:8001/sse`.

5

Does text_to_image return the actual image?

No, it returns only the URL of the generated image. Use the `download_image` tool or access the URL in a browser to obtain the actual image.

评论

其他 分类下的更多 MCP 服务器

Awesome Mlops

visenger

A curated list of references for MLOps

14.0K

DeepMark

DeepMark helps teachers deliver rapid, consistent marking with meaningful feedback for every student — in a fraction of the time. What once took a week, now takes one free period.

Blender

ahujasid

Open-source MCP to use Blender with any LLM

23.6K

EverArt

modelcontextprotocol

Model Context Protocol Servers

88.1K

Volley.sh

Volley

A good sequence diagram can solve almost any communication problem in engineering. Volley lets your agent write them as plain text, with the real request and response payloads on every arrow. It checks its own syntax wi

广告合作