Runs

A run is one execution attempt of a trigger.dev task. Every BGO an agent spawns, every scheduled task, every batch item produces at least one run.

Platos uses trigger.dev as the durable execution engine. Run is a trigger.dev concept that Platos surfaces in the dashboard for visibility. The semantics (queueing, retries, attempts, cancellation, machine selection, log capture) are unchanged from upstream.

In Platos

The Runs page at /orgs/{org}/projects/{project}/env/{env}/runs lists every run for the active environment. Filter by status, task identifier, or time window. Click a run to see its log stream, span timeline, and result.

When you click into a Platos task (BGO), the BGO detail surface links to the underlying run. That is the single click between "the agent dispatched this BGO" and "here is the engine-level execution record". Span events from the run feed Traces so the BGO timeline composes with the originating turn.

For "stuck" runs, see the Recover a stuck run recipe; the most common cause is a worker pool sized to zero or a queue paused at the engine layer.

Reference

The full reference for runs lives in the trigger.dev docs:

trigger.dev/docs/v3/runs

Platos tasks: the Platos surface that spawns runs.
Schedules: scheduled triggers create runs.
Queues: the queue a run waits in before execution.
Deployments: a run executes against a specific deployment of a task.
Traces: turn timelines compose run spans.