Agent Ops / Standard term
Trigger
A trigger is the event that causes an agent run to start. It can be time-based (a set hour), event-based (a new file or message arrives), or condition-based (a tracked number crosses a threshold).
A trigger is the event that causes an agent run to start, and it falls into one of three kinds: time-based, event-based, or condition-based. A time-based trigger fires on the clock, such as every weekday at 7 in the morning. An event-based trigger fires when something specific happens, such as a new file landing in a shared folder. A condition-based trigger fires when a tracked number crosses a line you set. Say you want a digest of unread mail once the inbox passes 50 messages: you tell the agent to watch that count and start the run the moment it crosses 50, so the work begins on its own without you opening anything.
Builder example
The trigger you pick decides how reliable and how easy to debug your agent is. A time-based trigger is the steady starting point because it runs on a predictable clock, so when output looks wrong you know exactly when it ran. Event-based and condition-based triggers react faster, yet they depend on watching an outside system, which adds places for a missed start or a double start. Tell your AI assistant the exact moment a run should begin, and start with a fixed time before you graduate to event or condition triggers.
Common confusion: A trigger decides when a run starts; a schedule decides how often a time-based run repeats. Every scheduled agent has a trigger, but not every trigger is a clock. Event-based and condition-based triggers start a run with no fixed time at all, firing only when their watched thing happens.