iTerm2 Agent FAQs

Question 1

What are the requirements to use iTerm2 Agent?

Accepted Answer

You need a macOS system with iTerm2 installed and its Python API enabled (via Preferences → General → Magic). Additionally, Python 3.11+ and `uv` for dependency management are required for installation and operation.

Question 2

What specific actions can I perform with iTerm2 Agent?

Accepted Answer

You can read visible terminal screen content and cursor position, execute shell commands and capture their output, send raw text or control characters (like Ctrl+C), monitor output for specific regex patterns, and manage iTerm2 sessions (list, create, split, close, focus panes).

Question 3

How does iTerm2 Agent integrate with AI tools like Claude Code?

Accepted Answer

iTerm2 Agent integrates as an MCP (Model Context Protocol) server. Once registered in your AI client's configuration (e.g., `~/.claude.json`), AI agents can directly invoke its tools (e.g., `run_command`, `read_screen`) to interact with and automate your iTerm2 terminals.

Question 4

What is iTerm2 Agent and what does it do?

Accepted Answer

iTerm2 Agent is an MCP server that enables AI agents and scripts to programmatically control iTerm2 terminal sessions on macOS. It allows for reading screen content, executing commands, sending keystrokes, monitoring output, and managing panes to automate complex terminal workflows.

Question 5

Are there any security considerations when running commands via iTerm2 Agent?

Accepted Answer

Yes, iTerm2 Agent classifies commands into 'SAFE', 'CAUTION', and 'DANGEROUS' categories. This security guard helps users, and particularly AI agents, understand the potential impact of commands, providing warnings for modifying or destructive operations like `rm` or `sudo`.

iTerm2 Agent

iTerm2 Agent

关于

主要功能

使用案例

关于

主要功能

使用案例