Operational issues

When something fails that a person must handle, Core opens an Agent Operational Issue. Each issue names what failed, how often, the run or approval it concerns, and its Next Step.

Only operators and admins see issues.

Find them

Open the Agent Operational Issues tab and choose the Open list view. The Agent Core Health card also counts them under Review open issues.

How issues behave

• The same failure on the same subject adds to one open issue instead of opening a new one.
• Core closes an issue itself when it sees the failure recover, for example a notice that later goes through.
• Close the rest yourself once you handle them, with the Dismiss Agent Operational Issue Flow action and a reason.
• Closed issues are deleted after 30 days.

Common issues

Code	Means	Do this
FLOW_START_LOST	A Flow started a run, but the start never happened. One issue per start, naming the agent and the record.	Start the agent again for the record, with the same Correlation Key if one was used.
FLOW_START_FAILED	A Flow’s start of a run failed.	Fix the reason in Detail, such as the agent’s sharing or its execution identity, then start the run again.
WORKER_FAILED	A step’s job failed.	Nothing at first. Recovery retries the step. If the run paused, rescue or cancel it.
WORKER_FAILURE_RECOVERY_FAILED	Recovery could not handle a failed step.	Click Run Recovery Now on the Health card. If the run paused, rescue or cancel it.
DISPATCH_RETRY_FAILED	Core could not start a run’s next step as the execution user.	Click Verify on the Agent Execution Identity. Recovery tries again within 10 minutes.
RECOVERY_ITEM_FAILED	Recovery failed on one item and skips it for 15 minutes.	If it keeps failing, open the run and rescue or cancel it.

Any other issue says: open the related run or record, fix the cause, then dismiss the issue.