# Add Agent Run

Field | Description
--- | ---
Channel | Where the run is triggered from.
Maximum iterations | The most steps the run is allowed to take.
Actions per iteration | The most actions the agent can take in a single iteration.
Continue autonomously until complete | Let the agent keep going without stopping for approval at each step, up to the iteration limit.

Once started, a run can be:

**Name** | **Description**
:--- | ---
Viewed | Open the run to see its progress and the actions it took.
Canceled | Stop a run that's still in progress.
Rolled back | Undo a completed run's changes.

!!! Note
Review a run's actions the same way you'd review any administrative change before trusting it. Rollback exists for a reason — use it if a run did something unexpected.
!!!