Terminal and command line

Everything the Studio does, you can also do from a terminal — and a few things (setup, the background service, scripting) start there. This page is the everyday tour; every command and flag is in the command-line reference. aivyx-pa --help lists them too, and aivyx-pa <command> --help explains one.

Three ways to chat from a terminal

Plain chat — just run:

aivyx-pa

Type a message and press Enter; the reply streams in as it’s written. Press Ctrl-C to stop a reply, and Ctrl-D to leave. When the assistant wants to do something irreversible it shows what and why, and asks Approve? [y/N] — type y to let it, anything else to refuse. This is also the one to use when piping text in from another program.

The terminal UI — a full-screen app with a chat pane and status bar:

aivyx-pa tui

Key	What it does
Enter	Send
Esc or Ctrl-C	Stop the reply in progress
PgUp / PgDn, ↑ / ↓	Scroll
Tab, or 1–5	Switch view: 1 Chat, 2 Missions, 3 Dashboard, 4 Audit, 5 Tools
`y` / `n`	Approve or refuse when an approval appears
`n` (Missions view)	Start a new mission
`a` / `r` (Missions view)	Approve or reject a mission waiting on you
Ctrl-Q	Quit (the assistant keeps running)

One task from a script — no conversation, just the answer:

aivyx-pa --headless "Summarise today's calendar"
echo "Tidy the notes folder" | aivyx-pa --headless

Nobody is there to approve anything, so a headless task refuses irreversible steps instead of waiting. It needs the background assistant already running (aivyx-pa daemon run, or the installed service). The exit code tells a script what happened: 0 done, 3 refused at an approval (or it wanted cloud consent), 1 anything else. Good for cron jobs and shell scripts.

All three talk to the same background assistant, so a conversation in one shares memory with the others. If it isn’t running yet, aivyx-pa and aivyx-pa tui start it for you (when your passphrase is stored).

Commands inside a chat

These work in plain chat, the terminal UI, the Studio and the chat apps:

Command	What it does
`/models`	Which models routing can choose from
`/models why`	Why it chose the model it did
`/model <name>` / `/model auto`	Pin this conversation to a model, or let routing choose again
`/allow-cloud`	Let this conversation use a cloud model (see Models and routing)

The commands you’ll use most

Command	What it does
`aivyx-pa init`	First-time setup, or change the model, provider or keys
`aivyx-pa doctor`	Check everything and say what to fix
`aivyx-pa studio`	Print the Studio sign-in link
`aivyx-pa daemon status` / `stop` / `run`	Check, stop or start the background assistant
`aivyx-pa daemon install`	Run it as a background service that starts at login
`aivyx-pa access show` / `set <level>`	See or change how far it can reach
`aivyx-pa autonomy show` / `set <level>`	See or change how much it does on its own
`aivyx-pa memory`	Browse, search and tidy its memory
`aivyx-pa learning`	What it has learned recently
`aivyx-pa skills` / `persona` / `profile`	Its skills, character and your Profile
`aivyx-pa team`	Run and manage team missions
`aivyx-pa loop`	Stock and run the autonomous backlog
`aivyx-pa cost`	What your model use has cost
`aivyx-pa connect`	Connect Gmail, Calendar, Drive or Contacts
`aivyx-pa instances`	List, create or remove separate assistants

Roles and instances on the command line

--role <name> starts a conversation as one of your roles; aivyx-pa --print-role shows what the active role can do.
--instance <name> talks to one of your named instances instead of the default assistant. It works with every command.

Keeping it running

aivyx-pa daemon install sets up a background service — a systemd user service on Linux, a launch agent on macOS — so your assistant starts when you log in, runs routines on time and keeps the Studio available. On Linux, read its log with journalctl --user -u aivyx-pa-daemon -f. aivyx-pa daemon uninstall removes it again.