Scripting & agents

The same predictable flow that guides a human drives cleanly from scripts, CI, and AI agents. Any of --license, --set, or --get-tokens switches off the prompts and runs as a single command, and the standalone modes (--verify, --apply-config, --remove-headers) never prompt at all.

Exit codes are the contract

Every run signals its outcome through its exit code, so a script or agent always knows what happened without parsing output:

• A successful generation, a clean --verify, or a healed run exits 0.
• A missing required field, an unrecognized identifier, a --verify --strict run that found drift, or a standalone mode with nothing to act on exits non-zero — and prints exactly what's wrong.

Standalone modes take priority

--apply-config and --remove-headers are standalone: when combined with selection flags like --license, the standalone mode wins. This means an agent can replay a saved config without worrying that a stray --license will override it.

Tell your agent to use License Wizard and it will: the flags are explicit, the failures are descriptive, and nothing is written until every required field is present.