Monitor Runs
Review workflow run history, understand run statuses, read stage-by-stage results, and diagnose errors.
Prerequisites
- At least one completed or in-progress workflow run (see Run a Workflow)
Steps
1. Open the run history
Navigate to your workflow and click the Runs & Triggers tab. The run history table shows all runs for this workflow, newest first.
2. Understand run statuses
Each run has a status badge:
| Status | Badge color | Meaning |
|---|---|---|
| Pending | Gray | Run created but not yet started |
| Running | Blue | Workflow is actively executing |
| Completed | Green | All stages finished successfully |
| Stopped | Yellow | Manually stopped by a user |
| Error | Red | A stage or agent encountered a failure |
3. Review run metadata
Each row in the history table shows:
- Run ID -- Unique identifier for the run
- Status -- Current state badge
- Started at -- When the run began
- Completed at -- When it finished (if applicable)
- Duration -- Total elapsed time
- Triggered by -- Whether started manually or by a trigger
4. Expand a run to see events
Click a run row to expand it and view the full event timeline. Events are listed chronologically and show exactly what happened at each step of the workflow.
5. Read stage outputs
For each completed stage, the event timeline includes the stage's output. Look for Stage completed events to read what each agent produced.
If a stage's output is missing from the timeline, check its summary mode. Stages with summary mode set to "none" do not include their output in the event stream.
6. Diagnose errors
When a run has Error status, expand it and look for the error event. The event includes:
- Which stage failed
- Which agent was executing
- Error message from the engine
- Tool call details if a tool invocation caused the failure
Common error causes:
| Error | Likely cause | Fix |
|---|---|---|
| Authentication failed | Invalid or expired API key | Update the credential in Settings |
| Tool not found | Tool removed or renamed | Update the skill or agent tool assignments |
| Rate limit exceeded | Too many API calls | Wait and retry, or switch to a different LLM |
| Timeout | Stage exceeded max cycles | Increase max cycles or simplify the task |
| Runtime unavailable | Code execution runtime not running | Start or reconfigure the runtime |
7. Compare runs
Review multiple runs of the same workflow to identify patterns. Consistent failures at the same stage suggest a configuration issue. Intermittent failures may indicate rate limiting or external service instability.
8. Filter by project
To see runs across all workflows in a project, navigate to the project's overview page. The recent runs section shows a consolidated view of all workflow activity.
9. Clean up old runs
Run history persists indefinitely. Each run stores its events and metadata for future reference. There is no automatic cleanup -- all runs remain available for review.
What's next
- Run a Workflow to start a new run
- Schedule Workflows to automate execution
- Configure Hooks to add pre/post processing that may prevent errors