Agent Automations → History is the complete record of your agent’s work: every run of every Agent Automation, searchable and paginated, newest first.

​
What each row tells you

  • The run: which Agent Automation fired and when
  • The outcome: succeeded, failed, cancelled, or skipped
  • The cost: what the run spent in credits
Open any row for that run’s chat, the step-by-step account of what the agent actually did. That chat is live: reply to it and the agent continues with the run’s full context.

​
Reading the outcomes

  • Succeeded needs no reading
  • Failed means the run didn’t finish; its chat shows where it stopped. A common cause is a run that couldn’t fit its interval: a job on a five-minute cadence gets five minutes, then is stopped so the schedule can’t back up
  • Cancelled means someone pressed stop, from the run’s chat or to take the agent back for a conversation
  • Skipped means the trigger fired while the agent was busy, and the row names what was holding it. Skipped runs never execute and never spend

​
Using it day to day

The history is where trust gets built. Skim it weekly and you know your agent’s rhythm: what ran, what it cost, and whether any schedule wants slowing down.