Shell MCP
Securely execute shell commands with whitelisting, resource limits, and timeout controls for LLMs.
shell-mcp
Shell command execution MCP server
An MCP server implementation that provides secure shell command execution capabilities for LLMs.
Features
- Secure Execution: Whitelisted commands and arguments only
- Resource Control: Memory and CPU usage monitoring
- Timeout Control: Automatic termination of long-running commands
- Size Limits: Output size restrictions for safety
- MCP Integration: Works with Claude and other MCP-compatible LLMs
Installation
npm install @kevinwatt/shell-mcp
Configuration with Dive Desktop
- Click "+ Add MCP Server" in Dive Desktop
- Copy and paste this configuration:
{
"mcpServers": {
"shell": {
"command": "npx",
"args": [
"-y",
"@kevinwatt/shell-mcp"
]
}
}
}
Tool Documentation
- shell_ls, shell_pwd, shell_df, etc.
- Execute whitelisted shell commands
- Inputs:
command(string, required): Command to executeargs(array, optional): Command argumentstimeout(number, optional): Execution timeout in ms
Usage Examples
Ask your LLM to:
"Show current directory using shell_pwd"
"List files using shell_ls with -l argument"
"Check disk usage using shell_df with -h argument"
Manual Start
If needed, start the server manually:
npx @kevinwatt/shell-mcp
Requirements
- Node.js 18+
- MCP-compatible LLM service
Development
# Install dependencies
npm install
# Watch mode
npm run watch
# Run tests
npm test
# Lint
npm run lint
License
MIT © Dewei Yen
Keywords
- mcp
- shell
- command
- claude
- llm
- automation
Available Commands
The following shell commands are available:
| Command | Description | Allowed Arguments |
|---|---|---|
| ls | List directory contents | -l, -a, -h, -R, --help, * |
| cat | Concatenate and display file contents | -n, -b, --help, * |
| pwd | Show current working directory | None |
| df | Show disk usage | -h, -T, --help |
| echo | Display text | Any text |
| ps | Show process status | -e, -f, -u, --help |
| free | Show memory usage | -h, -m, -g, --help |
| uptime | Show system uptime | None |
| date | Show system date and time | +%Y-%m-%d, +%H:%M:%S, --help |
| grep | Search text patterns in files | -i, -v, -n, -r, -l, --color, * |
| w | Show who is logged on and what they are doing | -h, -s, --no-header, --help |
| whois | Query WHOIS domain registration information | -H, * |
| find | Search for files in a directory hierarchy | -name, -type, -size, -mtime, * |
| netstat | Network connection information | -a, -n, -t, -u, -l, -p, --help |
| lspci | List PCI devices | -v, -k, -mm, -nn, --help |
| lsusb | List USB devices | -v, -t, -d, -s, --help |
| dig | DNS lookup utility | +short, +trace, +dnssec, @*, * |
| nslookup | Query DNS records | -type=, -query=, * |
| ip | Show network devices and interfaces | addr, link, route, neigh, -br, * |
| whereis | Locate binary, source and manual files | -b, -m, -s, * |
Related Servers
MCP RAG Server
A lightweight Python server for Retrieval-Augmented Generation (RAG) using AWS Lambda. It retrieves knowledge from external data sources like arXiv and PubMed.
GraphQL API Explorer
Provides intelligent introspection and exploration capabilities for any GraphQL API.
Code Analysis MCP Server
A modular MCP server for code analysis, supporting file operations, code search, and structure analysis.
Figma MCP Server
Connects to the Figma API, allowing AI tools to access and interact with your Figma designs.
Chronos Protocol
A robust MCP server that eliminates temporal blindness in AI coding agents through intelligent time tracking, persistent memory, and complete session traceability.
Random Number
Provides LLMs with essential random generation abilities, built entirely on Python's standard library.
Python MCP Server for Code Graph Extraction
Extracts and analyzes Python code structures, focusing on import/export relationships.
Playwright IA: Midscene.js
Automate Playwright flows using natural language with Midscene.js and AI.
mcpGW
An MCP gateway and proxy designed to simplify applications that utilize various tools.
ZeroPath MCP Server
Interact with your product security findings using natural language.