iFlytek Workflow MCP Server

Overview

iFlytek Workflow MCP Server enables you to connect to iFlytek's AI workflow platform through the Model Context Protocol. This allows you to create and execute complex AI workflows with features like sequential execution, parallel processing, loops, and nested workflows.

Installation

To install the iFlytek Workflow MCP Server, add the following configuration to your MCP client's configuration file (claude_desktop_config.json or mcp.json):

{
    "mcpServers": {
        "ifly-workflow-mcp-server": {
            "command": "uvx",
            "args": [
                "--from",
                "git+https://github.com/iflytek/ifly-workflow-mcp-server",
                "ifly_workflow_mcp_server"
            ],
            "env": {
                "CONFIG_PATH": "/path/to/your/config.yaml"
            }
        }
    }
}

Replace /path/to/your/config.yaml with the actual path to your configuration file.

Configuration

Before using the MCP server, you need to create a config.yaml file with your workflow information. The file should follow this format:

- flow_id: 'your_flow_id'              # required
  name: 'your_flow_name'               # optional
  description: 'flow description'       # optional
  api_key: 'API Key:API Secret'         # required

You can add multiple workflows by adding more entries to this list.

Getting Workflow Authentication Information

To use iFlytek workflows, you'll need to:

1. Create a bot on the iFlytek platform
2. Debug your workflow to ensure it works correctly
3. Publish your workflow as an API
4. Bind the workflow to an application
5. Retrieve the workflow ID and authentication information

If you don't have an application to bind to, you can apply for one at https://www.xfyun.cn.

Workflow Capabilities

iFlytek workflows offer several powerful features:

• 14 node types: Support for basic, tool, logic, and transformation nodes
• Complex variable I/O: Flexible data transmission between nodes
• Multiple execution modes: Sequential, parallel, loop, and nested execution
• Streaming output: Real-time processing through hook mechanisms
• Multiple development paradigms: From simple linear flows to complex multi-turn interactions
• Multi-model support: Based on the Model of Models (MoM) architecture for flexible model combinations

Once your MCP server is configured, you can interact with your iFlytek workflows directly through your MCP-compatible applications.