AI-Shell-Team/aish

What it solves

AISH is an AI-native interactive shell designed to bridge the gap between natural language requests and system operations. It allows users to perform complex terminal tasks—such as troubleshooting, file management, and system monitoring—by describing them in plain English rather than manually typing complex shell commands.

How it works

It functions as a full PTY (Pseudo-Terminal) shell, meaning it can run interactive programs like vim, ssh, and top just like a standard terminal. When a user starts a command with a semicolon (;), the shell enters AI mode. The system then uses a configurable LLM to generate, explain, and execute the requested command. To ensure safety, it includes a security policy engine that grades the risk of AI-generated commands (Low, Medium, High) and can optionally run commands in a sandbox to assess changes before they are applied to the actual system.

Who it’s for

System administrators, DevOps engineers, and power users who want to automate operations and troubleshoot systems more efficiently without leaving the terminal environment.

Highlights

• Full PTY Support: Unlike some AI CLI tools, it supports fully interactive terminal applications.
• Risk-Based Security: Implements a confirmation flow for medium-risk commands and blocks high-risk ones by default.
• ** uma Sandbox Pre-run**: Optional capability to test AI commands in a sandbox before production execution.
• Extensible Skills: A plugin system that supports hot-loading to expand the AI's domain knowledge.
• Model Agnostic: Supports a wide range of models via OpenAI-compatible APIs and local LLMs.