Glossary

Plain-language definitions of the terms this bundle uses. It grows as the bundle does.

• Playbook — a procedure your AI assistant reads and runs on its own, start to finish, rather than a single prompt you paste.
• Agent-executable — written so an AI assistant can follow the steps directly, not just so a person can understand them.
• Connector (tool) — the link that lets your assistant read a service such as your email or calendar. Also called a tool or an integration.
• Read-first — the assistant reads and drafts, but does not send, delete, or change anything on its own.
• Trigger — when the playbook is meant to run: on a schedule (each morning) or when you ask.
• Guardrails — the limits the assistant must respect while running a playbook, such as never sending mail or acting on anything ambiguous.