Testing and running
The run tab on a step, manual start with test data, failure reasons, and Recent runs.
The run tab
Every step has a small ▶ run tab. Pressing it starts the published flow from that step, as if the steps before it had just completed. It is the fastest way to try a Then branch without waiting for the trigger.
If nothing happens, the tab shows why:
| Reason | What to do |
|---|---|
| Publish the flow first | the run tab executes the published graph; a draft-only flow has none |
| Flow is disabled | switch it on in the header or the list |
| On cooldown | wait out the cooldown, or lower it in the header menu |
| Already running | the hub's concurrent-run limit is reached; wait for a run to finish |
| Not connected | the hub is offline or restarting |
Running by hand needs only the run automations permission, so household members can trigger a flow they cannot edit.
Manual start with test data
The Manual start When step carries test values: a device to pretend triggered, mock values
for trigger.* paths (you can type the label, Open, or the raw token, on), and preset variables. They are sent from your working copy, so you can
tweak them and run without publishing — the graph that runs is still the published one.
Recent runs
From the Flows list, Recent runs shows the last runs of a flow with the trigger, the status, and any Log step messages (capped per run). A run that ended at a Stop step still shows as successful, with where it stopped.
Flows also appear in the location's shared automation run history next to smart apps.
Changing a flow that is running
Publishing aborts in-flight runs of that flow, including parked waits. Pending waits in other flows are untouched.