Claude Code is an AI coding agent that lives in your terminal. You give it a goal in plain language; it plans, runs commands, reads and edits files, and reports back — the same loop a human engineer runs, automated.
| Mode | Command | When |
|---|---|---|
| Interactive console | claude | Working on a task together, back and forth |
| Headless (print mode) | claude -p "prompt" | Scripts, pipelines, one-shot answers |
Headless mode is the one that composes with everything else on Linux: its answer goes to stdout, so you can pipe it, redirect it, or call it from cron.
Your VM has Claude Code preinstalled and authenticated against the lab's AI
gateway with your personal budgeted key. The model is pinned by environment
variables — claude-opus-4-8 by default, with sonnet and haiku aliases for
lighter work. Nothing to configure.
By default Claude asks before it touches your files or runs commands; in
headless mode you grant that up front with --permission-mode acceptEdits
(file edits allowed) — you stay in charge of everything else.
Next: verify those prerequisites on this machine.