Agent Ops / Standard term
Cron schedule
A cron schedule runs an agent at fixed clock times regardless of what changed since the last run, such as every morning at 7 or every Monday at 9. The run time is fixed in advance, so the agent wakes on a predictable beat.
A cron schedule runs an agent at fixed clock times regardless of what changed since the last run, such as every morning at 7 or every Monday at 9. You give the agent a pattern (the day, the hour, the minute) and it wakes up on that beat whether or not there is new work to do. Say you want a digest of yesterday's open tasks waiting each morning: you tell the assistant to run at 6:30 every weekday, pull the task list, and write the summary to a note. The schedule decides when; the agent decides what to do once it wakes. The name comes from the long-standing Unix cron utility that pioneered this clock-based pattern.
Builder example
Schedules are how you turn a one-off prompt into a standing routine the agent runs without you. If you want a weekly review email, a nightly backup summary, or a Monday planning note, a cron schedule fires it on time every time. The risk is firing on empty: a 7 a.m. run still costs a model call and writes an output even on a day with nothing new, so pair it with a quick check that skips or shortens the work when there is nothing to report.
Common confusion: A cron schedule fires on the clock; an event trigger fires when something happens, like a new file arriving. What separates them is the cause: time versus an event. A clock-based run is predictable but can wake to find no work, while an event-based run only starts when there is something to react to.