awesome-remote-mcp-servers

A Remote Model Context Protocol (MCP) server is a service that enables AI models to interact with external tools and resources over the internet. Unlike local MCP servers that require installing NPM packages, remote servers can be accessed via a URL, making them compatible with web-based AI clients (like the web versions of Claude or ChatGPT) and easier to deploy.

Key benefits include:

• Ease of use: Connect by simply copying and pasting a server URL.
• Web Compatibility: The only way to use MCP servers with web-based clients.
• Security: Servers can be verified via their domain and use standard authentication methods like OAuth 2.1 or API Keys.

Remote MCP servers typically use one of two endpoint patterns:

• /sse: Uses the Server Sent Events (SSE) protocol. This is being slowly deprecated.
• /mcp: Uses the Streamed HTTP protocol. This is the preferred modern standard.

Future clients may support self-discovery by using the prefix preceding 'sse' or 'mcp' to find the full URL.

1. Go to Settings → Connectors → Advanced Settings and enable Developer Mode.
2. In connector settings, click create.
3. Fill in the following:
   • Name: (e.g., test)
   • URL: (e.g., https://mcp.my-awesome-server.com/sse)
   • Authentication: OAuth
4. In a new chat, ensure Developer Mode is turned on and the connector is selected.

Note: Developer Mode must be enabled and this feature may not be available to all users.

Add the server configuration to your Windsurf MCP configuration file.

{
  "mcpServers": {
    "test": {
      "serverUrl": "https://mcp.my-awesome-server.com/sse"
    }
  }
}

Navigate to the MCP Servers section → Remote Servers → Edit Configuration and add the following configuration:

{
  "mcpServers": {
    "test": {
      "url": "https://mcp.my-awesome-server.com/sse",
      "type": "streamableHttp"
    }
  }
}

Remote MCP servers listed in this repository can be consumed in two primary ways:

1. Using an MCP-ready client: Connect directly through interfaces like Claude, ChatGPT, or Cursor. This typically involves providing the server URL to the client.
2. In an API request to an LLM provider: Integrate the server capabilities when making programmatic calls to LLM providers such as OpenAI, Anthropic, or Gemini.

Run the following command in your terminal to add a remote MCP server using the HTTP transport:

claude mcp add --transport http "test" https://mcp.my-awesome-server.com/sse

Add the server configuration to your ~/.gemini/settings.json file.

{
  "mcpServers": {
    "test": {
      "httpUrl": "https://mcp.my-awesome-server.com/sse"
    }
  }
}

You can use an OAuth-protected MCP server in API requests to LLM providers (Anthropic, OpenAI, Gemini), but you must manually manage the OAuth authentication flow to obtain an access token.

For testing purposes, you can follow Anthropic's guide on Obtaining an access token for testing. However, manual token management is not recommended for production use as it is not a secure way to authenticate users.

Navigate to Settings → Connectors → Add Custom Connector and provide the following details:

• Name: (e.g., test)
• URL: (e.g., https://mcp.my-awesome-server.com/sse)

Note: If you are part of an organization, you may need administrator permission to access custom connectors.

To use a remote MCP server in Cursor, add the server configuration to your ~/.cursor/mcp.json file (global) or .cursor/mcp.json (project-specific).

{
  "mcpServers": {
    "test": {
      "url": "https://mcp.my-awesome-server.com/sse"
    }
  }
}

To use a remote MCP server in VS Code, add the server configuration to your settings.json file.

{
  "mcp": {
    "servers": {
      "test": {
        "type": "http",
        "url": "https://mcp.my-awesome-server.com/sse"
      }
    }
  }
}