広告掲載
MCP.so
ログイン

MCP Server Commands

@g0t4

リポジトリ

基本情報

カテゴリ

その他

ライセンス

MIT

トランスポート

stdio

作者

g0t4

MCP Server Commands について

Model Context Protocol server to run commands (tool: `runProcess`)

MCP Server Commands の設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "mcp-server-commands": {
      "command": "uvx",
      "args": [
        "mcpo",
        "--port",
        "3010",
        "--api-key",
        "supersecret",
        "--",
        "npx",
        "mcp-server-commands"
      ]
    }
  }
}

MCP Server Commands のツール

1

run a command, i.e. `hostname` or `ls -al` or `echo "hello world"` etc

概要

What is Mcp Server Commands?

Mcp Server Commands is an MCP server that provides a single tool runProcess to run commands on the host machine. It supports both shell command lines (with pipes, redirects, variable expansion) and direct executable invocation via an argv array. The server returns standard output and standard error as text, and optionally accepts stdin. It is designed for use with LLMs that need to execute system commands and use the output in conversations.

How to use Mcp Server Commands?

Install the npm package mcp-server-commands and configure it in your MCP client (e.g., Claude Desktop). Invoke the runProcess tool with either a command_line string (shell interpreted) or an argv array (direct executable). Optionally provide a stdin string to pass data to the command’s standard input. The tool returns STDOUT and STDERR as text. For local models, use a system prompt that encourages tool usage.

Key features of Mcp Server Commands

  • Runs commands via shell or direct argv invocation.
  • Returns stdout and stderr as separate text outputs.
  • Supports optional stdin for passing scripts or input.
  • Single tool design, easy for models to learn.
  • Security warning: always review commands before approving.
  • Works with Claude Desktop, Groq Desktop, and HTTP via mcpo.

Use cases of Mcp Server Commands

  • Run simple commands like hostname or ls -al and feed output to the model.
  • Execute scripts passed via stdin to interpreters (bash, python, fish, etc.).
  • Create files from model-generated content using cat >> path with stdin.
  • Enable an LLM to execute user‑requested system commands interactively.

よくある質問

1

How is the command executed?

Two mutually exclusive ways: `command_line` (string) runs through the default shell with pipes, redirects, and variable expansion; `argv` (array) executes the executable directly without shell interpretation.

2

How can I pass input to a command?

Use the optional `stdin` parameter. The text you provide will be fed to the command's standard input – useful for scripting or creating files with `cat >>`.

3

What are the security considerations?

The server runs with the user's permissions. Do not run with `sudo`. In Claude Desktop, use "Approve Once" to review each command; deny if you don't trust it.

4

How do I install Mcp Server Commands?

Published as `mcp-server-commands` on npm. Add the config to `claude_desktop_config.json` with `command: "npx"` and `args: ["mcp-server-commands"]`. Or build locally from the repository with `npm run build` and point to the built script.

5

How can I debug the server?

Use the MCP Inspector via `npm run inspector`. Logs are written to STDERR; add `--verbose` to the args for more messages. Claude Desktop logs to `~/Library/Logs/Claude/mcp-server-mcp-server-commands.log`.

コメント

「その他」の他のコンテンツ

Blender

ahujasid

Open-source MCP to use Blender with any LLM

23.6K

EverArt

modelcontextprotocol

Model Context Protocol Servers

88.1K

🚀 Model Context Protocol (MCP) Curriculum for Beginners

microsoft

This open-source curriculum introduces the fundamentals of Model Context Protocol (MCP) through real-world, cross-language examples in .NET, Java, TypeScript, JavaScript, Rust and Python. Designed for developers, it focuses on practical techniques for building modular, scalable,

16.7K

Volley.sh

Volley

A good sequence diagram can solve almost any communication problem in engineering. Volley lets your agent write them as plain text, with the real request and response payloads on every arrow. It checks its own syntax wi

YOUB

Connect your AI assistant to your own YOUB endurance training account. Read recorded activities with full analysis, planned sessions, race goals, wellness (HRV, sleep, resting HR) and your athletic profile. Create or mov

広告掲載