open source

Type it.
Run it.
That simple.

Command is an AI-powered terminal. Write what you want in plain language — it translates to shell commands and executes them. Local or cloud, your pick.

command — bash
// why command

Less memorizing.
More doing.

No scripting, no man pages, no stack overflow. Just describe what you need.

$ natural language

Plain English → Shell

Describe any task. Command calls the AI, extracts commands from the response, and runs them — one by one.

$ offline mode

100% Local with Ollama

Run entirely offline. No API key. No data leaving your machine. Pull a model once and use it forever, free.

$ cloud mode

OpenAI Support

Prefer the cloud? Set your OPENAI_API_KEY and pick from GPT-4o-mini to GPT-5-nano.

$ bypass

Direct Execution

Prefix with ! to skip the AI and run any shell command raw. Full control, zero friction.

$ model switch

Swap Models Live

Type : to pick interactively, or :modelname to switch instantly — no restart needed.

$ safety guard

Destructive Block

Built-in blocklist stops catastrophic commands like rm -rf/ from ever being executed.

// get started

One line to install.

The installer sets up dependencies, configuration, and PATH — then asks whether you want local AI or cloud.

Windows powershell

Run in PowerShell. Admin rights recommended.

powershell -ExecutionPolicy Bypass -c "irm https://aiu4.com/command/install.ps1 | iex"
macOS / Linux bash

Works on macOS, Ubuntu, Debian, Arch, and most distros.

curl -fsSL https://aiu4.com/command/install.sh | bash
Prefer manual setup? Read the README →
// ai models

Local or cloud.

Switch models at any time without restarting. The installer walks you through the first setup.

qwen2.5:3b
Default. Best for code and commands.
~2 GB
gemma3:1b
Ideal for low-RAM machines.
~800 MB
gemma3:4b
Better for reasoning tasks.
~3 GB
llama3.2
General purpose use.
~2 GB
mistral
Advanced general tasks.
~4 GB