# Triggers

A pipeline starts by hand unless it declares a trigger. `schedule` runs it on a cron.
`webhook` runs it when an HTTP request arrives at a receiver path. `watch` looks at a
source on a cadence: a registered dataset, an S3 prefix, or a database table past a
watermark, and proposes or starts the pipeline when something changed. A fresh
install arms nothing: a declared schedule or watch starts paused until someone turns
it on.
