A simple implementation of an MCP server for using dify workflows, enabling the invocation of Dify workflow through MCP tools.
153 stars1 watching19 forks

Model Context Protocol (MCP) Server for dify workflows

A simple implementation of an MCP server for using dify. It achieves the invocation of the Dify workflow by calling the tools of MCP.

🔨Installation

Prepare config.yaml

Before using the mcp server, you should prepare a config.yaml to save your dify_base_url and dify_sks. The example config like this:

dify_base_url: "https://cloud.dify.ai/v1"
dify_app_sks:
  - "app-sk1"
  - "app-sk2"

Different SKs correspond to different dify workflows.

Run mcp server

Then you can run the dify mcp server in your clients. The config of client should like the following format:

"mcpServers": {
  "mcp-server-rag-web-browser": {
    "command": "uv",
      "args": [
        "--directory", "${DIFY_MCP_SERVER_PATH}",
        "run", "dify_mcp_server"
      ],
    "env": {
       "CONFIG_PATH": "$CONFIG_PATH"
    }
  }
}

Example config:

"mcpServers": {
  "mcp-server-rag-web-browser": {
    "command": "uv",
      "args": [
        "--directory", "/Users/lyx/Downloads/dify-mcp-server",
        "run", "dify_mcp_server"
      ],
    "env": {
       "CONFIG_PATH": "/Users/lyx/Downloads/config.yaml"
    }
  }
}

Enjoy it

At last, you can use dify tools in any client who supports mcp.

Features

workflows
configuration
integration
tools

Category

Integrations