Automate → Runs
Every execution of a playbook produces a run. This tab lists them all, with live status updates as new runs start, progress, and complete. Click a run to open its detailed replay.
What you can do
- Filter runs by playbook, triggering event, and status.
- See runs update live as they happen.
- Open a run to inspect step inputs and outputs, and why each Switch selected its output.
Filters
All filters apply automatically and stack.
| Filter | Notes |
|---|---|
| Playbook | Show only runs of a specific playbook. |
| Event | Show only runs triggered by a specific event type. |
| Status | Pending, Waiting for approval, Running, Success, Failed, Suppressed. |
Suppressed deserves a word: when a playbook has a cooldown configured (see Studio), a trigger that fires again within the cooldown window still creates a run — but the run is recorded as Suppressed and no actions execute. It is your visible proof that the throttle worked.
Table columns
| Column | Notes |
|---|---|
| Run | Link into the run replay, labeled {Playbook name} #{run id}. |
| Resource | For runs triggered by resource-scoped events, the resource involved (linked to its detail page). |
| Event | The event that triggered this run. |
| Status | Color-coded badge: blue Running, green Success, red Failed, amber Waiting for approval or Suppressed, grey otherwise. |
| Started | Timestamp the run started. |
| Finished | Timestamp the run finished, or - while still in flight. |
| Actions | An eye button that opens the run replay (same as clicking the run name). |
A Live badge at the top of the tab confirms that updates are streamed in real time — no manual refresh needed.
Opening a run
Click the linked run name (or the eye button). The Automation Replay page opens, showing:
- The playbook graph as it was when the run executed, with node borders and status labels reflecting running/retrying, waiting for approval, success, failure, or skipped. The connections actually followed are highlighted. Alternative branches may reconnect to a shared continuation; if that continuation ran, it retains its real status. Unvisited steps become Skipped when the run finishes.
- An Inspector panel on the right. Click the trigger to see the source event data and the configured filters; click an action node to see its status, attempt count, start/finish times, its Bindings and Configuration, the Resolved Input it actually received, and its Output; click a connection to see whether it was traversed during this run.
An Approve control pauses its run as Waiting for approval. Users with app.automation.write can cancel a waiting run while its request is undecided. See Approvals for decisions, expiry, and recovery. If an action fails and has retries configured, the retries happen automatically during the run; the Inspector shows the attempt count.
Inspecting a Switch
Select a Switch to see its selected named rule, Fallback, or Error, and the evidence for that choice. Rules and conditions show Match, No match, Error, or Not evaluated, alongside condition sources and bounded value previews. Later rules are not evaluated after the first match; conditions skipped by All/Any short-circuiting also show Not evaluated. Arrays and objects show summaries rather than every item.
Labels and expected values come from the run’s saved playbook snapshot. Renaming or reordering rules in today’s playbook does not change an earlier run. If a Switch’s historical evidence is unavailable, the Inspector shows Routing details were not recorded.
Fallback means every evaluated rule was false and the Switch succeeded. Error means an encountered operand was missing, null when a comparison requires a value, of the wrong type, or an invalid date. Evaluation stops at that error without checking later rules. Switch conditions are not retried.
A Switch forwards incoming data unchanged, including success: false from an earlier failed action. Its own status describes whether routing evaluated successfully, so a successful Switch can display a forwarded failure output. An unconnected named route or Fallback preserves the incoming run outcome. An unconnected Error ends failed; a connected Error can be handled by a successful subsequent action, but a chain of Switch controls alone does not erase it. See Switch control for route and collection comparisons.
Inspecting an approval
Select an Approve control to see its recorded title and message, approver roles, deadline, decision, deciding user, and comment. The Inspector can also show eligible-user counts and email delivery details. Continuation queued means a decision is saved but the next execution segment has not started. Waiting runs have no completion timestamp and remain available during normal run-history cleanup.
An unconnected Approved output completes successfully; unconnected Denied and Expired outputs finish as Failed. A successful action on a connected branch can handle denial or expiry. Cancelling an undecided waiting run skips the remaining steps and does not follow an output. See Approve control for eligibility, configuration, and decisions.
The saved request and decision remain available after the playbook is edited or deleted, while their history is retained. You can also inspect automation approval evidence in Review → Approvals.
Permissions
- Read runs and open replays:
app.automation.read. - Cancel an undecided waiting run:
app.automation.write. - Deciding an approval requires the selected approver role and applicable resource-read scope; permission to inspect a run does not grant approval eligibility.