Agent activity logs

8 min

The Agent Activity Logs tab unifies execution activity for specialized agents and workflow agents in Optimizely Opal. Monitor recent executions, troubleshoot failures, and cancel runaway executions in one place.

Access the Agent Activity Logs tab

To open the Agent Activity Logs tab from the Agents page, go to Agents > Agent Activity Logs.

screenshot of the Agents page where the Agent Activity Logs tab is selected

To open the Agent Activity Logs filtered to a specific workflow agent, click More > View Logs for that agent. See Workflow agent logs for details.

Filter executions

The Agent Activity Logs tab includes filters that narrow the list of executions in the table.

screenshot of the Agent Activity Logs tab where the Agent Type, Status, Triggered By, and Date Range filters are visible

Agent Type filter – Filter executions by agent type.

Status filter – Filter executions by status. See Execution statuses for descriptions of each status.

Triggered By filter – Filter executions by the user who triggered them.

Date Range filter – Filter executions by timeframe.

Search by agent name

Enter an agent name in the Search by agent name field. Results update as you type.

Configure columns

Show, hide, and reorder columns in the table to control which information displays.

Click Configure Columns.

screenshot of the Configure Columns panel where column toggles are visible

Toggle a column on or off.

Drag a column to reorder it.

Table columns

The table includes the following columns:

Type – An icon indicating the agent type (specialized or workflow).

Execution Time – The date and time the execution started.

Agent – The agent name and ID.

Trigger – The trigger type that started the workflow execution. See Workflow agent triggers for descriptions of each trigger type.

Triggered By – The user who triggered the execution.

Status – The current execution status. See Execution statuses for descriptions.

Credits – The approximate credit usage for the execution.

Evaluation Score – The quality score for the execution. This column displays a value only for executions of agents that have preferred output examples.

Duration – The total time the execution took to run.

Private executions

When an agent has Private executions enabled, its executions still display in the table, but only the user who triggered each run can open the contents.

A lock icon marks every private execution.

Private executions that you did not trigger are greyed out, and you cannot open them. The tooltip reads This is a private execution — only its owner can open it.

Your own private executions display normally, with the lock icon.

See Create a specialized agent to enable private executions for an agent.

Cancel multiple executions

Cancel multiple executions at once instead of stopping each one individually. This is useful when a workflow triggers more executions than expected or when many executions queue up.

Select the checkbox for each execution to cancel. To select all executions, select the checkbox in the column header.

Click Stop executions.

screenshot of the Agent Activity Logs table where executions are selected and the Stop executions option is visible

Click Cancel Executions to confirm.

Opal cancels only executions with Running or Pending statuses. When you cancel, any step in progress finishes and no further steps start.

Execution statuses

The Status column displays the current state of each execution. Each status badge also has a reason that explains what happened and why. Hover over a status badge to see its reason. Examples include waiting for execution capacity and a user stopping the execution.

Use the reason to find the cause of an unexpected result and decide what to do next, without guessing or contacting support.

screenshot of the Status column where the pointer rests on a status badge and its reason displays

The following table lists each status, its reasons, and what each reason means. Descriptions that end with "(workflows only)" apply only to workflow agents.

Status	Reason	What it means
Pending	Queued	The execution is waiting in line to start.
Pending	Waiting for a free slot	All capacity is busy. The execution starts when a slot opens.
Pending	Deferred for priority	The execution gave up its slot so higher-priority work could go first.
Pending	Rate limited	Too many executions started recently. The execution waits until the limit resets.
Running	Started	The execution is working.
Running	Resumed	The execution continued after a retry, a pause, or an answer to its question.
Paused	Waiting on input	The execution asked a question or is waiting on an external service.
Completed	Success	The execution finished normally.
Failed	Timeout	The execution took too long and was cut off.
Failed	Error	The execution hit an error it could not recover from.
Failed	Validation failed	The inputs did not pass checks, so nothing ran.
Failed	Output rejected	The result was reviewed and judged unusable (workflows only).
Failed	Invalid setup	The workflow configuration is invalid, so the workflow never started (workflows only).
Stopped	Unusual behavior detected	The system halted the execution before it finished because its behavior looked anomalous.
Stopped	No tools available	The system blocked every tool the execution tried to use, so the execution could not continue.
Stopped	Blocked by content moderation	Content moderation flagged the content, so the system never sent it to the model.
Stopped	Blocked by input rules	The agent's own rules rejected the request.
Stopped	Security threat detected	A scan flagged a file for suspicious content.
Stopped	No matching path	A condition checked every option, none matched, and there was no fallback (workflows only).
Stopped	Too many iterations	The execution hit the maximum allowed iterations (workflows only).
Stopped	Too many steps	The execution hit the maximum allowed steps (workflows only).
Stopped	Gave up	The workflow planner, the component that decides which step runs next, determined that it could not make progress (workflows only).
Cancelled	Stopped manually	A person stopped the execution.
Cancelled	Disconnected	The browser tab closed or the connection dropped during the execution.
Superseded	Replaced by newer run	A newer message in the same conversation took over.
Superseded	Replaced by retry	The system retried the failed execution and abandoned this one.
Superseded	Replaced by resume	The system resumed a cancelled execution and abandoned this one.
Unknown	Not recorded	The system did not capture the outcome.

Specialized agent logs – View detailed logs for specialized agent executions.

Workflow agent logs – View detailed logs for workflow agent executions.

Note

If you use Opti ID, administrators can turn off generative AI in the Opti ID Admin Center. See Turn generative AI off across Optimizely applicationsTurn generative AI off across Optimizely applications.