​
CLI Usage

The AIUS CLI is a TUI (Terminal User Interface) for autonomous data-science workflows. The agent loop runs server-side; the CLI renders the run, executes tools locally, and pauses at review gates for your input.

​
Installation

Self-contained binaries for macOS, Linux, and Windows: 
curl -fsSL https://aius.co/install.sh | bash
See the Installation guide for details and options.

​
Authentication

On first launch, the TUI offers three ways to sign in:
  • Log in — with your email and password (handles email verification and 2FA if enabled)
  • Create account — register a new account without leaving the terminal
  • Paste API key — use an existing aius_… token from your dashboard
Whichever you choose, the CLI stores a token at ~/.local/share/aius/auth.json with mode 0600 — your password itself is never saved. The same flows are available as commands: 
aius auth login                  # email + password prompt
aius auth login --key aius_xxx   # use an API key directly
aius auth register               # create an account from the terminal
You can also provide a key via environment variable: 
export AIUS_API_KEY=your_aius_key
You don’t supply an LLM provider key — model traffic is routed through the AIUS service, which holds those credentials server-side.

​
Running a project

AIUS works on a project directory containing your brief and dataset: 
my-project/
  context/   # your brief (markdown, docs)
  data/      # your dataset (CSV, parquet, …)
Launch it from inside the project: 
aius
…or point at a directory: 
aius /path/to/my-project
AIUS reads the brief, profiles the data, and walks the analysis pipeline — pausing at review gates so you can Continue or steer it.

​
Common options

aius --continue              # resume the last session
aius --session <id>          # resume a specific session
aius --model <provider/id>   # override the model for this run
aius --reset                 # wipe local session state (auth is preserved)
aius --log-level DEBUG       # raise log verbosity
aius --print-logs            # stream logs to the terminal

​
Agent capabilities

The agent can:
  • Read — analyze files and understand project structure
  • Edit / Write — modify and create files (notebooks, CONTEXT.md, models, dashboards)
  • Execute — run commands and notebooks within the project, gated by approval

​
Security

  • Actions that touch your machine (shell commands, file writes/edits, access outside the project) are gated by permission prompts before they run.
  • Your AIUS key is stored locally with restricted file permissions (0600).
  • The CLI only sends what’s needed to drive the run to the AIUS service.

​
Requirements

  • macOS, Linux, or Windows
  • An AIUS account — sign in from the TUI with email and password, or with an API key from your dashboard