Make_MCP

What is Make_MCP?

Make_MCP is an MCP server designed to help users create new MCP servers. It provides documentation and example code so that Claude can generate a basic MCP server script on demand.

How to use Make_MCP?

Add Make_MCP to Claude Desktop by editing the claude_desktop_config.json file with a command pointing to uvx (or a local Python installation). After restarting Claude, enable the make_mcp server’s how_to_make_mcp tool, then ask Claude to write a new server (e.g., “Write a basic MCP server with a simple URL fetch tool”). Alternatively, attach the server’s resource manually via the “+” menu.

Key features of Make_MCP

• Provides a how_to_make_mcp tool with documentation and example code
• Offers a resource that attaches the same documentation to a message
• Generates a new MCP server script as a downloadable artifact
• Integrates seamlessly with Claude Desktop’s tool and resource menus
• Supports both uv and manual Python installation

Use cases of Make_MCP

• Quickly scaffolding a new MCP server from scratch
• Learning how MCP servers are structured and configured
• Demonstrating the Tool and Resource interaction patterns in Claude Desktop

FAQ from Make_MCP

How do I install Make_MCP?

You can use uv with the provided JSON configuration, or clone the repository and point the command to your Python interpreter with args set to the path of make_mcp/core.py after installing mcp.

What is the difference between the tool and resource methods?

The tool method lets Claude call how_to_make_mcp when needed to receive documentation. The resource method lets you manually attach the same documentation to your message, but the tool method is recommended.

How do I add a new server created by Make_MCP to Claude Desktop?

After downloading the generated script, add a new entry to your claude_desktop_config.json with the appropriate command and arguments, restart Claude Desktop, and enable the new server’s tools.

Does Make_MCP require any specific runtime?

It requires either uv (recommended) or a Python environment with the mcp package installed and accessible to your configured Python interpreter.