Triggers: what starts a run

A run has to be started by something, and on YAMO there are four things that can start one. Whichever it was, the run itself is identical: the same readiness checks, the same order of steps, the same charging, the same approval rules. A trigger changes who asked, not what happens.

Someone confirms it

The ordinary way. You open an Instance of a Service, answer what it asks, and confirm. The conversation shows you the exact price if one applies and waits for you to agree before anything starts. Runs execute in the background, so closing the tab changes nothing.

A schedule fires

A Playbook whose document declares a cadence runs on it, without anyone present. Scheduled runs are ordinary runs: same readiness checks, same charging, same approval gates. A schedule never bypasses a gate. See Schedules.

An approved agent calls it

A published Service can be marked agent callable, which lets an approved external agent start a run over YAMO's programmatic endpoint on behalf of whoever granted it consent. That consent carries a spend cap, and revoking it cuts the agent off. See the MCP endpoint.

A paused run resumes

A run held at an approval gate continues from exactly where it stopped once somebody decides. It picks up its own saved state rather than starting again, and it does not quietly adopt a newer version of the Playbook halfway through. See Approvals.

What cannot start a run

An external event cannot. A Playbook can declare that it wants to run on one, and the document will parse, but there is nothing that can send that event yet, so deploying such a Playbook is refused with that reason rather than accepted and left silent.

A Draft Service cannot be run by anything, including you. Running starts at Deploy. See What is a Service?.