Remote MCP Server on Cloudflare

This document provides detailed instructions and technical guidance to set up, deploy, and integrate an MCP (Model Context Protocol) server on Cloudflare Workers, with a focus on enabling seamless AI application integration.

Overview: What is Remote MCP Server?

The Remote MCP Server enables the connection of AI applications like Claude Desktop, Continue, Cursor, and other compatible clients to various data sources and tools. By leveraging Model Context Protocol (MCP), this server serves as a bridge for these AI apps to perform operations such as calculations, database queries, or any other contextual tasks.

🔧 Core Features & MCP Capabilities

The Remote MCP Server offers several key features that empower integration with different AI applications:

• OAuth Authentication: Securely enables users to authenticate and access services.
• Real-Time Communication: Supports real-time interaction through Server-Sent Events (SSE).
• Tool Execution: Facilitates the execution of specified tools or functions within an AI application.

These capabilities are essential for developers looking to build robust AI workflows, ensuring seamless interaction between their applications and external data sources or tools.

⚙️ MCP Architecture & Protocol Implementation

The architecture of the Remote MCP Server is designed to adhere strictly to the Model Context Protocol (MCP). This includes:

• Protocol Compliance: Ensures compatibility with various MCP clients.
• Worker-based Deployment: Utilizes Cloudflare Workers for efficient and scalable deployment.

Mermaid Diagram: MCP Protocol Flow

graph TB
    A[AI Application] -->|MCP Client| B[MCP Protocol]
    B --> C[MCP Server]
    C --> D[Data Source/Tool]
    style A fill:#e1f5fe
    style C fill:#f3e5f5
    style D fill:#e8f5e8

Mermaid Diagram: Data Architecture

graph TB
    A[Persistent Storage] --> B[MCP Server Cache]
    B --> C[SSE Stream]
    C --> D[AI Application]
    E[MPC Client] --> F[MCP Server Endpoint]
    style A fill:#f5ede8
    style C fill:#d6e9c7

🚀 Getting Started with Installation

To get started, follow these steps to set up and deploy the Remote MCP Server on Cloudflare Workers.

Local Development Environment Setup

1. Clone the Repository

   git clone [email protected]:cloudflare/ai.git
   

2. Install Dependencies

   cd ai
   npm install
   

3. Run Locally

   npx nx dev remote-mcp-server
   

You should be able to open http://localhost:8787/ in your browser.

Connecting the MCP Inspector

To explore and interact with the API:

• Start the Inspector

  npx @modelcontextprotocol/inspector@latest
  

• Switch Transport Type to SSE: Set URL as http://localhost:8787/sse.
• Login: Enter a dummy email/pass for the login screen and proceed.

Integrating with Claude Desktop

1. Modify Configuration File

   {
     "mcpServers": {
       "math": {
         "command": "npx",
         "args": [
           "mcp-remote",
           "http://localhost:8787/sse"
         ]
       }
     }
   }
   

2. Connect to Remote Server Replace the local URL with the worker-url when deploying remotely.

💡 Key Use Cases in AI Workflows

1. Real-Time Data Processing: Integrate real-time data processing tools within an AI application.

   • Example Scenario: Stream live financial market data and process signals using MCP.

2. Interactive User Interface: Enable interactive prompts for users to call specific tools or services.

   • Example Scenario: Allow users to input parameters into mathematical calculations on-the-fly through MCP.

🔌 Integration with MCP Clients

MCP Client	Resources	Tools	Prompts
Claude Desktop	✅	✅	✅
Continue	✅	✅	✅
Cursor	❌	✅	❌

📊 Performance & Compatibility Matrix

The Remote MCP Server is optimized for:

• Cloudflare Workers Deployment: Ensures fast and reliable performance.
• OAuth2 Authentication: Secure and user-friendly authentication process.

🛠️ Advanced Configuration & Security

Configuration Example

{
  "mcpServers": {
    "[server-name]": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-[name]"],
      "env": {
        "API_KEY": "your-api-key"
      }
    }
  }
}

Debugging and Maintenance

• Restarting Claude: Restart the application to resolve issues.
• Clear Auth Files: Clear session storage with rm -rf ~/.mcp-auth.

❓ Frequently Asked Questions (FAQ)

1. Why does My MCP Client Not Work?

   • Ensure you have correctly configured your client and server.

2. How Do I Secure My Integration?

   • Use OAuth2 for secure authentication.

3. Can I Deploy on a Different Platform?

   • Yes, but ensure compliance with the Model Context Protocol (MCP).

4. What Tools are Compatible?

   • Most AI applications that support MCP clients will be compatible.

5. How Do I Debug MCP Server Issues?

   • Clear session storage and restart both the server and client.

👨‍💻 Development & Contribution Guidelines

Contributions to this project are welcome! Please follow these guidelines:

• Ensure compatibility with various MCP clients.
• Follow existing coding standards for consistency.

🌐 MCP Ecosystem & Resources

Explore more about Model Context Protocol (MCP) in the official documentation and community forums. Additionally, stay updated on the latest developments by following the Cloudflare Worker blog.

By leveraging the Remote MCP Server, developers can enhance their AI applications with powerful tool integration, ensuring a seamless user experience across multiple platforms and services.