Trigger
Trigger
Also called: 触发器 · Event trigger
The event source that starts one agent run. It can be a user message, a timer firing, a webhook from an external system, or another agent's output.
Common types
- User trigger: a message, a slash command, a button click.
- Time trigger: a cron expression firing (see "Scheduled agent").
- Event trigger: a new GitHub PR, an incoming email, a monitoring alert — usually delivered via webhook.
- Agent trigger: an upstream agent finishes and passes its result as the downstream agent's starting input.
The trigger decides "when to start" and often carries the initial input of "what to handle this time."
Design considerations
When triggers come from outside, consider: deduplication (the same event delivered twice), rate limiting (a burst of triggers), and authentication (confirming the event source is trusted). These are runtime responsibilities and should not be left to the model.
Common misconceptions
- A trigger only "starts" the run; it does not decide what the agent does afterwards — that is prompt and process.
- An unauthenticated external webhook trigger is an attack surface; the runtime must verify the source.
Related terms
Sources
Compiled 2026-08-29 · This glossary is compiled from public papers, official specifications, and common industry definitions, and is updated as the field evolves. Corrections welcome.