Monitoring Runs and Output Quality
The Agent Jobs dashboard gives you a unified view of every run across all your services. Use the filter bar to narrow results by service, run status, or date range — useful when you need to triage failures or audit output volume for a specific client.
Run Status Reference#
| Name | Type | Required | Description |
|---|---|---|---|
pending | status | No | Job has been created and is waiting to be picked up. No agent activity yet. |
running | status | No | Agent is actively executing. Output is streamed in real time on the run detail page. |
completed | status | No | Run finished successfully. Full output is available and ready to review or deliver. |
failed | status | No | An error stopped the run before completion. The error message is shown in run detail; a manual retry is available. |
aborted | status | No | Run was cancelled mid-execution, either by the operator or by a timeout policy. |
Diagnosing a Failed Run#
Open the run detail and read the error message
Click the failed run from the Agent Jobs dashboard to open its detail page. The error message is displayed at the top of the panel — read it in full before taking any action.
Check the intake data that was sent
Expand the intake JSON section on the run detail page. Verify that all required context fields were present and correctly formatted. A missing or malformed field is the most common cause of agent failure.
Verify the OpenClaw agent is running
Open the OpenClaw dashboard and confirm the agent definition linked to this service is active and healthy. If the agent is paused or errored at the infrastructure level, the run will always fail regardless of intake data.
Re-run with the same data or fix the intake form first
If the intake data was correct and the agent is healthy, use the Retry button to re-run with the same intake payload. If the intake data was the problem, fix the intake form or ask the client to resubmit, then retry.
Quality Signals#
After an order is delivered, the client can leave a thumbs up or thumbs down rating directly from their portal. This rating is recorded against the run and visible in the run detail panel, giving you a per-delivery quality signal without any back-and-forth.
Reviewing Output#
The full output text is displayed in a scrollable panel on the run detail page. Use the copy to clipboard button to grab the complete output in one click — useful for pasting into a client report, a CMS, or a review document.