Skip to main content

Worklist

Worklist is the asynchronous operations layer for long-running and scheduled tasks.

How to get there: Agent → Worklist (when enabled)


What Worklist is for​

Use Worklist to:

  • run background tasks outside live chat turns,
  • schedule recurring operations,
  • monitor status and outputs,
  • rerun or archive failed/completed tasks.

Task list structure​

ColumnMeaning
NameTask title
StatusCurrent lifecycle state
OutputGenerated result/summary
ScheduleRecurrence details
DescriptionTask intent/context
CreatedCreation timestamp
Chat ThreadSource conversation link
ActionsTrigger/rerun/edit/archive/delete

Task list overview


Task statuses​

StatusInterpretation
pendingQueued and waiting
in_progressCurrently executing
needs_reviewCompleted but needs operator review
completedFinished successfully
failedExecution failed

Actions and operator controls​

ActionUse case
TriggerStart queued task now
RerunRetry with same setup
EditAdjust task details before next run
ArchiveRemove from active queue view
DeletePermanently remove task

worklist edit and delete


Recurrence scheduling​

Worklist supports recurring jobs with configurable cadence.

SettingPurpose
FrequencyDaily/weekly/custom cadence
IntervalEvery N minutes/hours/days
Days of weekWeekly targeting
TimezoneScheduling context
End conditionStop policy (date/count/never)

worklist recurring


Debugging failed tasks​

When tasks fail:

  1. Open task details/output
  2. Check related thread context
  3. Validate tool configuration and credentials
  4. Validate file input assumptions
  5. Apply fix and rerun

Worklist + chat relationship​

Tasks often originate from user chat actions but may execute later.

Use Chat Thread link to connect:

  • original user request,
  • generated task config,
  • execution result.

Best practices​

  • Use recurring tasks for deterministic routine workflows
  • Keep task descriptions explicit for operators
  • Review failed/needs_review queue daily
  • Archive completed tasks to keep active view clean
  • Define rerun policy for high-impact tasks