Studio

You can turn any CLI into an AI-friendly MCP tool with the Studio MCP server. It lets your AI agent call your existing commands as MCP endpoints, making it easy to integrate patching CLIs, debugging MCPs, or running custom scripts without building a full MCP from scratch.

How to use

Use an MCP client to call the Studio MCP server and invoke your CLI commands as MCP tools. You write a studio command that describes the inputs your tool expects, and Studio exposes those arguments to the AI as a structured tool. For example, you can turn a text-to-speech command into a spoken-output tool, or patch a translation CLI into an AI-enabled helper.

How to install

Prerequisites you need to have on your system:

• Node.js and npm installed on your machine
• A working terminal or shell

Install the Studio MCP server globally with npm:

npm install -g @studio-mcp/studio

Configuration samples

Use Studio to patch a CLI by turning its command into an MCP tool. The examples below show how to wire the Studio tool into three common client configurations.

npx -y @studio-mcp/studio command "{ required_argument # Description of argument }" "[optional_args... # any array of arguments]"

Configuration examples for common clients

Claude Desktop example configuration for a fixed say command using Studio (replace with your actual setup as needed). This defines a tool named say that invokes Studio via npx with your chosen arguments.

{
  "mcpServers": {
    "say": {
      "command": "npx",
      "args": [
        "-y",
        "@studio-mcp/studio",
        "say",
        "-v",
        "siri",
        "{speech # A concise message to say outloud}"
      ]
    }
  }
}

Cursor configuration example

Add the Studio-based MCP server to your Cursor configuration to enable calling your CLI through Cursor. Note that Cursor may have issues with arguments containing spaces in some versions.

{
  "mcpServers": {
    "say": {
      "command": "npx",
      "args": [
        "-y",
        "@studio-mcp/studio",
        "say",
        "-v",
        "siri",
        "{speech#say_outloud}"
      ]
    }
  }
}

VSCode configuration sample

In VSCode, you can wire a Studio-based MCP server similarly to patch an echo tool that prompts you for the text to send.

{
  "mcp": {
    "servers": {
      "echo": {
        "type": "stdio",
        "command": "npx",
        "args": [
          "-y",
          "@studio-mcp/studio",
          "echo",
          "{text#What do you want to say?}"
        ]
      }
    }
  }
}

Available tools

Studio MCP tool

Turns a CLI into an MCP tool by parsing the Studio command template and exposing a structured input schema for the AI to fill in.

MCP tool schema

Provides a JSON input schema that describes required and optional arguments for the tool so the AI can generate valid calls.

Template syntax

Uses a Mustache-like syntax to describe required and optional arguments and their descriptions for the MCP tool.