Overview

MCP Server Commands allows Large Language Models (LLMs) to execute commands and scripts on your local machine through the Model Context Protocol. This enables AI assistants like Claude to perform tasks directly on your system, from simple file operations to complex scripting.

Available Tools

run_command

This tool allows the LLM to execute a single command on your system, such as:

• ls -al to list files
• hostname to get your computer's name
• echo "hello world" to output text

The tool returns both STDOUT and STDERR as text, which the LLM can then analyze and respond to.

run_script

This more powerful tool allows the LLM to execute multi-line scripts using various interpreters:

• Shells: bash, fish, zsh
• Programming languages: python and others

The script content is passed over STDIN, making this essentially equivalent to run_command plus script input. This enables the LLM to write and execute code on the fly, creating a highly interactive experience.

Security Considerations

Important: Be careful what commands you allow this server to run! Commands will execute with the same permissions as the user running the server.

Best practices:

• Use the "Approve Once" option in Claude Desktop (not "Allow for This Chat") to review each command
• Use "Deny" if you don't trust a proposed command
• Never run the server with elevated privileges (e.g., sudo)
• Review command output carefully before continuing conversations

Installation

Using npm (Recommended)

The package is published to npm as mcp-server-commands. To install and configure:

1. Install the package globally or use npx:

   npm install -g mcp-server-commands
   

2. Configure Claude Desktop by editing the config file:

   • MacOS: ~/Library/Application Support/Claude/claude_desktop_config.json
   • Windows: %APPDATA%/Claude/claude_desktop_config.json

3. Add the following configuration:

   {
     "mcpServers": {
       "mcp-server-commands": {
         "command": "npx",
         "args": ["mcp-server-commands"]
       }
     }
   }
   

From Source

If you prefer to build from source:

1. Clone the repository:

   git clone https://github.com/g0t4/mcp-server-commands.git
   

2. Install dependencies:

   cd mcp-server-commands
   npm install
   

3. Build the server:

   npm run build
   

4. Configure Claude Desktop by editing the config file as above, but point to your local build:

   {
     "mcpServers": {
       "mcp-server-commands": {
         "command": "/path/to/mcp-server-commands/build/index.js"
       }
     }
   }
   

Debugging and Logging

• Claude Desktop writes logs to ~/Library/Logs/Claude/mcp-server-mcp-server-commands.log
• Add --verbose to the args in your configuration for more detailed logging
• For advanced debugging, use the MCP Inspector:

  npm run inspector
  

Usage Examples

Here are some examples of how to use the tools with Claude:

1. Simple file listing: Ask Claude to "Show me the files in my current directory"

2. System information: Ask Claude to "Tell me about my system configuration"

3. Create a file: Ask Claude to "Create a simple Python script that prints Hello World"

4. Run a data analysis: Ask Claude to "Analyze the contents of my CSV file using Python"

Remember to review all commands before approving them to ensure they're safe to execute on your system.