Workflow Runs
Each time you trigger a workflow, it creates an Workflow Run that tracks the execution from start to finish.
What is a Workflow Run?
A workflow run is a single execution of a workflow. It captures:
- When the run started and ended
- Success or failure status
- Step-by-step execution details
- Generated artifacts
- Any errors or warnings
Think of it as a complete audit trail for one workflow execution.
Triggering a Workflow Run
From the UI
- Navigate to your workflow
- Click Run Workflow
- Provide required inputs (if any)
- Click Start Run
The workflow will execute each step in sequence, and you'll see real-time progress.
Via API
You can trigger workflow runs programmatically via the Workflows API for automated workflows.
Learn more: Workflows API Documentation
Run Information
Each run displays the following information:
Basic Details
- Start time - When execution began
- End time - When execution completed (or current time if still running)
- Status - Success, failed, or in progress
- Duration - Total time from start to end
Execution Details
- Step-by-step log - See what happened at each step
- Generated artifacts - Links to created artifacts
- Input values - The inputs provided for this run
- Processing time - How long each step took
Status Indicators
| Status | Meaning |
|---|---|
| Pending | Run created, not yet started |
| Executing | Steps are being processed |
| Completed | All steps completed successfully, artifacts generated |
| Failed | One or more steps failed, see error details |
| Cancelling | Cancel has been requested |
| Cancelled | Run was manually stopped |
Note: Awaiting User Input is a step status, not a run status. While the overall run stays in Executing, an individual step can enter awaiting_user_input when it needs input from you before it can continue.
Viewing Workflow Runs
Access Run History
- Navigate to your workflow
- Click the Runs tab
- View list of all executions (most recent first)
Run List View
The runs list shows:
- Run name (editable — click to rename, auto-generated with timestamp by default)
- Start time
- Duration
- Status
- Generated artifacts count
Click any run to view full details.
Run Details View
Click on a specific run to see:
Steps Sidebar: A sidebar shows all steps in the workflow with visual status indicators. Each step displays its current status (pending, executing, completed, failed, etc.) so you can track progress at a glance.
Step-Level Chat: Each step has its own persistent chat interface. You can interact with the AI at the step level to provide inputs, ask questions about the step's output, or refine results.
Collapsible Execution Log: Detailed execution logs are available in a collapsible panel. Expand to see the full log of what happened during each step, including MCP tool calls and timing.
Artifacts Generated:
- Links to all artifacts created
- Artifact types and names
- Creation timestamps
Error Information (if failed):
- Which step failed
- Error message
- Stack trace (if applicable)
- Suggested fixes
Input Values:
- All workflow inputs provided
- Values used for each step
Editable Run Names
Run names are auto-generated with timestamps (e.g., "Run 2025-01-15 14:30"). You can click on the run name to rename it to something more meaningful.
Cancel Runs
You can cancel a run that is currently in progress:
- Open the run details
- Click the Cancel button
- The run transitions to Cancelling and then Cancelled status
Cancelled runs preserve any artifacts that were already generated before cancellation.
Monitoring Runs
Real-Time Monitoring
When a run is in progress, you'll see:
- Current step being executed
- Progress indicator
- MCP tool calls happening in real-time
- Partial outputs (where available)
Post-Execution Review
After completion, review:
- Success paths - Verify quality of generated artifacts
- Timing - Identify slow steps for optimization
- Tool usage - See which MCP tools were called
- Output quality - Check if results meet expectations
Understanding Run Logs
Each step in the run log shows:
Step 1: Gather Requirements
├─ Started: 2:30:15 PM
├─ MCP Tools Called:
│ └─ query_documents (3 queries)
├─ Duration: 12.3s
├─ Output: requirements_list (1,247 characters)
└─ Status: Success ✓
Step 2: Generate PRD
├─ Started: 2:30:27 PM
├─ Input: requirements_list from Step 1
├─ Duration: 34.6s
├─ Output: prd_draft (5,891 characters)
└─ Status: Success ✓
Step 3: Quality Check
├─ Started: 2:31:02 PM
├─ Input: prd_draft from Step 2
├─ Duration: 8.1s
├─ Artifact Generated: artifact-789 (PRD)
└─ Status: Success ✓
Common Run Issues
Run Failed at Step
What it means: One step in the workflow encountered an error
What to do:
- Check the error message for the failed step
- Verify the skill is valid
- Check if inputs were provided correctly
- Review MCP tool permissions (if tools were used)
Run Timed Out
What it means: The run exceeded the maximum execution time
What to do:
- Check if any step is making too many MCP tool calls
- Optimize prompts to be more efficient
- Consider breaking long workflows into multiple workflows
Artifacts Not Generated
What it means: Run completed but no artifacts were created
What to do:
- Verify artifact type and category settings
- Check project associations
- Ensure final step is configured to output an artifact
- Review step configurations
Run History and Retention
History Duration
- All runs are preserved indefinitely
- Access complete execution history anytime
- Export run data for analysis
Run Artifacts
- Generated artifacts persist even if run is deleted
- Artifact links remain valid
- Run history shows which run created each artifact
Analyzing Run Performance
Identify Bottlenecks
Look for steps that take significantly longer:
- Optimize prompts for those steps
- Reduce MCP tool calls if excessive
- Consider splitting complex steps
Success Rates
Monitor success vs. failure rates:
- High failure rate → Review prompt quality
- Intermittent failures → Check input validation
- Consistent failures → Verify configuration
Artifact Quality
Review generated artifacts from multiple runs:
- Consistent quality → Good skills
- Variable quality → Add more guardrails
- Poor quality → Refine prompts and validation
Best Practices
Review Early Runs
When first deploying a workflow:
- Monitor the first 5-10 runs closely
- Check artifact quality
- Verify timing is acceptable
- Look for any errors or warnings
Set Up Monitoring
For production workflows:
- Check runs regularly (daily or weekly)
- Set up alerts for failed runs
- Review artifact quality periodically
Learn from Failures
When runs fail:
- Read error messages carefully
- Check what inputs were provided
- Review the step that failed
- Update prompts or configuration as needed
Track Improvements
Keep notes on changes:
- Document prompt updates
- Record performance before/after changes
- Track success rates over time
Next Steps
- Create a workflow: Creating Workflows
- Trigger via API: Workflows API
- Learn about artifacts: Artifacts Overview