Model Context Protocol high risk

Commands

This MCP server provides a tool to run shell commands or executables, returning STDOUT and STDERR, enabling various system interactions but requiring careful usage due to potential risks.

Connections & Capabilities

Connects To

GitHub

Capabilities

execwrite

Quickstart

Install

npm install
```

Config

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

Exposed MCP Tools (1)

critical

run_process

Runs a specified command or executable and returns the standard output and standard error.

Allows execution of arbitrary commands, potentially leading to system compromise.

Safety Assessment

This server is safe when used with caution, reviewing each command before execution. It is risky when running untrusted commands or without understanding the potential consequences, especially due to the lack of sandboxing.

Requires user approval for each command in Claude Desktop.
Permissions are dictated by the user running the server.
Provides STDOUT and STDERR for command output, aiding in auditing.
Explicit mode selection (executable vs. shell) reduces ambiguity.

Can execute arbitrary commands, potentially leading to system compromise.
No built-in sandboxing; relies on OS-level permissions.
Potential for command injection if inputs are not carefully validated.
Server runs with the permissions of the user executing it, which can be elevated.