Model Context Protocol high risk

Iterm

iTerm-MCP allows models to interact with and control iTerm2, enabling automation and screen inspection, but requires careful monitoring due to potential risks.

Connections & Capabilities

Connects To

GitHub

Capabilities

readwriteexec

Quickstart

Install

npx -y @smithery/cli install iterm-mcp --client claude

Config

{
  "mcpServers": {
    "iterm-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "iterm-mcp"
      ]
    }
  }
}

Exposed MCP Tools (3)

high

write_to_terminal

Writes a command to the active iTerm terminal and executes it.

Unrestricted command execution can have significant consequences.

safe

read_terminal_output

Reads a specified number of lines from the active iTerm terminal.

Read-only operation with no side effects.

moderate

send_control_character

Sends a control character (e.g., Ctrl+C) to the active iTerm terminal.

Can interrupt processes or alter terminal behavior.

Safety Assessment

iTerm-MCP offers powerful terminal interaction capabilities but poses significant risks due to the lack of command restrictions and potential for unexpected model behavior. It's reasonably safe for experienced users who closely monitor model actions and understand terminal commands, but risky for novice users or unsupervised environments.

User is expected to monitor activity and abort when appropriate.
Focuses on terminal interaction, not direct system access.
Requires user to initiate and configure the MCP server.
Provides tools for reading and writing to the terminal, enabling interaction and automation.

No built-in restrictions on commands executed.
Models can behave unexpectedly, requiring constant monitoring.
Full terminal control allows potentially harmful commands.
Lacks sandboxing, increasing risk of unintended consequences.
No explicit authentication or authorization mechanisms.