Monitor Workflows & Agents
Track workflow runs, agent sessions, and review gates from Workflows, Work queue, agent details, or the linked work item.
Workflow overview
Open Workflows to see online agents, definitions, recent sessions, and recent runs. Open Work queue from the menu or title bar to inspect the active queue without leaving your page.
Queued work is not running yet. An eligible agent must claim the session before execution starts. Workflow runs and direct agent sessions share the same claim model.
Workflow runs
Workflow runs move through these statuses:
| Status | Meaning |
|---|---|
not_started | The run exists but has not entered the first step yet. |
running | An agent step is active or ready to be claimed. |
waiting_for_review | A human review step is waiting for a reviewer decision. |
stalled | The run needs intervention, such as a missing branch handoff. |
completed | The workflow reached a successful terminal route. |
cancelled | A user or system cancelled the run. |
failed | The workflow could not finish. |
Events record run creation, step entry, routing, agent start and completion, review requests, review decisions, review reassignment or escalation, resumption, cancellation, failure, and completion.
Agent sessions
Use Agent Sessions for the claim lifecycle behind a visible run. Session details show queued work, status, claim owner, activities, provider IDs, and the completion or failure state.
Archived sessions leave session lists, but the linked work item and its activity history stay.
Local agent status
Local agents send heartbeats while online. Without a heartbeat for 12 seconds, the platform marks the agent stale. After 5 minutes, it marks the agent offline, expires active claims, and affected sessions become stale.
From the CLI, one agent status shows background agents, local execution state, backpressure, and resource pressure. Foreground one agent watch prints progress in the current terminal.
Review gates
Human review steps wait in Work Inbox until a reviewer approves, sends feedback, rejects, or reassigns the gate.
Agent review steps must finish with one verdict: approve, feedback, or reject. A review-mode agent that completes without a verdict stalls the workflow instead of assuming success.