Peliqan offers an implementation of “Remote MCP”, allowing you to write and maintain your MCP Server functions (tools) in the cloud inside Peliqan, without the need for end-users to update their local installation when new functions are added.

This works using a Peliqan proxy, which is installed locally on the computer of the end-user, next to the MCP Client such as Claude Desktop or Visual Studio.

The actual MCP functions (tools) are defined in Peliqan in the cloud and can be updated at any time. The Proxy handles dynamic discovery of new functions (tools):

Compared to “classic” MCP, the Python functions (decorated with @mcp.tools()) are written and maintained in the cloud and not part of the local install on the end-user’s computer:

This means you can efficiently implement an MCP Server that provides a wide range of Data Features such as querying the data warehouse, running custom queries, making API calls and any custom logic that you need, all without local deployment hassle:

Below we walk you through the steps to setup your MCP Server using Peliqan’s Remote MCP.

Build a remote MCP Server in Peliqan

Add a custom API endpoint in Peliqan, and use the below script as the API handler.

Replace the functions decorated with @mcp.tools() with your own MCP functions (tools):

Using the Peliqan proxy (locally installed)

Below is an example Python script, that can be installed locally on the computer of the end-user - next to the MCP Client - and acts as a proxy.

Make sure to install required modules:

pip install mcp

pip install httpx

Configure your MCP Client - example for Claude Desktop

Example for Claude Desktop, create a file claude_desktop_config.json.

More info: