Yevanchen Difyapp As MCP Server

🚀 Exposing Dify Workflows as an MCP Server

This project exposes Dify workflows to AI clients supporting the Model Context Protocol (MCP), such as Claude Desktop and Cursor, through the MCP protocol on server ports.

🚀 Quick Start

This project implements a Dify plugin that allows you to expose Dify workflows to AI clients supporting the Model Context Protocol (MCP) via this protocol. With this plugin, you can:

• Offer Dify workflows as tools to AIs like Claude.
• Enable AI clients to discover and utilize your workflows.
• Integrate into AI assistants without modifying existing workflows.

✨ Features

• ✅ Support for the JSON - RPC interface of the MCP standard.
• ✅ Automatic tool discovery and registration.
• ✅ Compatibility with clients like Claude Desktop and Cursor.
• ✅ Secure implementation of SSE connections.
• ✅ Server - side session management.
• 🔄 Compliance with the latest MCP streaming HTTP specification (based on PR #206).

📚 Documentation

Background

The Model Context Protocol (MCP) is an open standard that enables AI models to interact with external tools and data sources. As MCP evolves, the protocol is transitioning from the HTTP + SSE mode to a new "streaming HTTP" transmission mode, providing better support for stateless servers.

This project tracks and implements the latest changes in the MCP protocol, specifically:

• The server is responsible for generating and managing session IDs.
• Support for the stateless server mode (suitable for the Dify plugin environment).
• Standardized message formats and stream processing.
• Secure session management and authentication.

Configuring Plugin Settings

Application ID:

Enter the Dify application ID you wish to expose to the MCP server.

Other Settings:

Configure other settings as needed.

Usage

1. Configure in Dify

Ensure that your Dify application contains at least one workflow and is correctly configured.

2. Use in the Client

1. Open the client.
2. Go to Settings > MCP Server.
3. Add a new MCP server and fill in the URL:

   https://your - Dify - instance - address/difyapp_as_mcp_server
   

4. Save and enable the server.

3. Use in Cursor

1. Open Cursor.
2. Go to Settings > AI > MCP.
3. Add the server address:

   https://your - Dify - instance - address/difyapp_as_mcp_server
   

4. Save and enable.
5. Use your tools in Cursor Agent.

🔧 Technical Details

Architecture

This plugin implements the MCP server through two endpoints:

• GET Endpoint: Handles SSE connections and HTML pages.
• POST Endpoint: Handles JSON - RPC requests.

Due to the limitations of the Dify plugin environment, we adopted a "minimum viable" SSE implementation, including:

• Server - side session ID generation.
• Limited heartbeat mode (approximately 5 minutes).
• Automatic reconnection after client disconnection.
• Compliance with the latest streaming HTTP specification.

Tool Registration

Tools are automatically generated from Dify workflows and exposed to clients via the MCP protocol. Tool definitions include:

• Name and description.
• Input parameter definitions.
• Return value types.
• Parameter validation.

Support for Stateless Mode

This plugin supports the stateless server mode compliant with the latest MCP specification, which means:

• The server does not need to maintain long - term connections.
• Each request is independent.
• Requests are associated through session IDs.
• It is suitable for the stateless API environment of Dify.

📚 Documentation

Troubleshooting

1. Connection Issues:
   • Ensure that the URL is correct and accessible.
   • Check if you are behind a firewall or proxy server.
2. Tools Invisible:
   • Ensure that the application ID is correctly configured.
   • Confirm that the Dify workflow is correctly configured.
3. Other Issues:
   • Check the plugin logs for more information.
   • Contact the support team for assistance.

Project Information

• Version: 1.0.0
• Author: [Your Name]
• Contact: [Your Contact Information]

By following the above steps, you can successfully expose Dify workflows as an MCP server to AI clients supporting the protocol.