Access a Remote MCP Server in Claude Desktop

If you’re using Claude Desktop to interact with AI assistants but want to leverage your own data or workflows, setting up a remote MCP server can completely change the game. This guide walks you through everything you need to know — from understanding what remote MCP servers are, to where you can find a list of publicly available ones, and how to configure Claude Desktop to use them.

What Is a Remote MCP Server?

MCP stands for Multi-Channel Processor (or sometimes Model Control Proxy, depending on context). In the case of Claude Desktop, it’s a middleware server that acts as a bridge between your Claude UI and backend services like:

• LLM APIs (e.g., OpenAI, Claude, DeepSeek, etc.)
• Data connectors (e.g., Salesforce, Google Sheets, SQL databases)
• Organizational tools (e.g., CRMs, file systems, analytics)

Remote MCP servers allow Claude Desktop to:

• Run custom pipelines and prompts.
• Query and summarize data from private sources.
• Interact with multiple services in parallel.
• Extend Claude’s capabilities without direct API coding.

Instead of running the MCP server locally, you can connect to a remote one, often hosted on a cloud VM (like AWS, Azure, or Lightsail).

Where to Find Public or Community Remote MCP Servers

A fantastic community-maintained list of remote MCP servers is available at: Awesome Remote MCP Servers GitHub Repository

This GitHub page is frequently updated and includes:

• Public access MCP servers (some allow anonymous use)
• Authenticated enterprise-grade servers
• Specialized servers with connectors for services like Salesforce, Notion, Xero, and more
• Installation scripts and Docker setups (for self-hosting)

You can choose one based on your use case or even contribute your own setup.

How to Configure Claude Desktop to Use a Remote MCP Server

❗ Before you begin: make sure you’ve installed the latest version of Claude Desktop. You can get it from claude.ai.

1. Install Claude Desktop (if not already installed)

• Visit: https://claude.ai
• Download the latest version for Windows, macOS, or Linux.
• Launch the app.

2. Locate Your Remote MCP Server URL

From the GitHub list above, pick a remote MCP server and copy its base URL. It should look like this:

https://mcp.monday.com/sse

Some servers might require authentication tokens or API keys. If so, make sure to note those.

3. Configure Claude Desktop

1. Open Claude Desktop

Make sure Claude Desktop is already installed and launched at least once so it can generate the config files.

2. Access the Configuration File

1. In Claude Desktop, click on:
   • File > Settings > Developer
2. Then click on the “Edit config” button.

This will open the configuration file named:

claude_desktop_config.json

3. Add Your MCP Server Entry

Here’s an example configuration for the Monday.com Remote MCP Server:

{
  "monday-api-mcp-hosted": {
    "command": "npx",
    "args": [
      "mcp-remote",
      "https://mcp.monday.com/sse"
    ]
  }
}

Explanation:
“monday-api-mcp-hosted” is a unique label you’ll see in the Claude MCP options.
npx mcp-remote https://mcp.monday.com/sse is the actual command Claude runs to create a bridge to the remote server.

4. Save and Restart Claude Desktop

Once you’ve saved the updated claude_desktop_config.json file:

• Close and restart Claude Desktop.
• On launch, Claude will detect the new MCP configuration.
• You may be prompted to authenticate with Monday.com (or whichever remote server you’ve set up).

4. Test the Connection

Once configured:

• Open a new Claude window.
• Ask something that requires MCP integration (e.g., “Pull my Salesforce contacts” or “Search Google Analytics for last month’s pageviews”).
• If the connection works, you’ll see a response powered by the remote MCP server.