Overview
The Workflow Run Execution page provides a real-time Kanban board view of your workflow execution. This page shows the status of each agent in your workflow as it runs, allowing you to monitor progress, review outputs, provide feedback, and manage the execution.
Accessing the Execution Page
The execution page automatically opens when you:
- Click "Run" from the Workflow Editor
- Navigate to a workflow run from your workflow history
- Access a scheduled workflow run
Kanban Board
The main area displays a six-column Kanban board showing agent statuses:
Column Structure
1. To Do (Gray)
- Agents that haven't started yet
- Waiting for dependencies to complete
2. Doing (Blue)
- Agents currently executing
- Shows real-time progress
3. Waiting (Amber/Yellow)
- Agents paused or waiting for conditions
- May be blocked by dependencies
4. Feedback (Purple)
- Agents requiring human input or validation
- Includes: FEEDBACK, AWAITING_VALIDATION, REVISE statuses
5. Done (Green)
- Successfully completed agents
- Outputs are available for review
6. Error (Red)
- Agents that failed during execution
- Error details available in the agent card modal
Each column header shows:
- Column name with status icon
- Count badge showing number of agents in that status
Agent Cards
Card Information
- Agent name and avatar
- Custom instructions or task description
- Visual status indicators through color and styling
Loop Cards
Special cards appear for loop logic blocks:
- Display a loop icon instead of agent avatar
- Show count of agents inside the loop
- Display avatars of agents within the loop with status dots
- Click to view detailed iteration information
Agent Output Modal
The Agent Output Modal provides complete execution details for an agent.
Header Section
- Agent Avatar: Large avatar image with blurred background
- Status Badge: Current execution status
- Run ID: Short identifier for this execution
- Agent Name: Title of the agent
- Metadata:
- Model name used
- Cost in credits (if available)
- Execution duration
- Start time
Main Content Area
1. Given Task
Shows the instructions or prompt given to the agent
Expandable if content is long
File references are simplified for readability
2. Thinking Process
Collapsible section showing real-time execution logs
Displays the agent's reasoning and decision-making process
For loops: Shows logs organized by iteration cycles
Updates automatically while agent is running
Shows "Monitoring execution..." when active
3. Agent Response
The final output from the agent
Formatted text or JSON display
Copy button to copy output to clipboard
Expandable for long responses
Shows error messages if execution failed
Shows "Working on the task..." animation when running
4. Loop Iteration Details (Loops Only)
Detailed breakdown of each loop cycle
Shows status of each agent within each iteration
Organized by iteration number
Sidebar Information
Generated Files
- List of all files created by this agent
- File name, size, and download options
- Preview button for viewing files
- Download all as ZIP option
- Files are organized in agent-specific folders
Media Gallery
- Grid view of images and videos generated
- Click to preview full-size
- Only appears if media files exist
Skills
- List of tools/capabilities the agent used
- Memory indicator if memory is enabled
- Human-in-the-loop indicator if applicable
Telemetry
- Initialization Time: When the agent started
- Execution Time: Duration of execution
- Status: Current status
- Global ID: Unique identifier for this run
Feedback Modal
When an agent requires feedback, a specialized modal appears.
Purpose
Agents with "Human in the Loop" enabled or those requiring validation will pause and request feedback before proceeding.
Modal Contents
1. Given Task
Shows what the agent was asked to do
Expandable for long instructions
2. Agent Response
Displays the output the agent generated
Expandable for long responses
Copy button available
3. Generated Artifacts
Media Gallery: Images/videos in a grid
- Hover to see preview and delete options
- Click to preview full-size
File List: Other generated files
- Preview and download options
- Delete option to exclude from feedback
4. Feedback Controls
Initial View:
Two buttons:
- Accept (Green): Approves the output and continues workflow
- Feedback (Red): Opens feedback form
Feedback Form:
Text area for describing needed changes
"Send Feedback" button to submit
Agent will revise based on feedback
Status Meanings
Workflow Statuses
- RUNNING: Workflow is actively executing
- DONE/COMPLETED: All agents completed successfully
- FAILED: Workflow encountered an error
- CANCELLED: Execution was stopped manually
- PAUSED: Workflow is temporarily paused
Agent Statuses
- TODO: Not started, waiting for dependencies
- DOING/RUNNING: Currently executing
- WAITING: Paused or waiting for conditions
- FEEDBACK/AWAITING_VALIDATION: Requires human input
- REVISE: Needs revision based on feedback
- DONE/COMPLETED: Successfully finished
- ERROR/FAILED: Execution failed
Real-Time Updates
Live Polling
- The page automatically refreshes every 2 seconds while workflow is running
- "Live" indicator shows when updates are active
- Agent cards move between columns as status changes
- Modals update automatically if kept open
Visual Indicators
- Pulsing dots: Active execution
- Status colors: Quick visual status recognition
- Count badges: Number of agents per status
- Connection lines: Show dependencies between agents (on To Do cards)
Managing Execution
Stopping a Workflow
1. Click "Stop" button in the header
2. Workflow will cancel gracefully
3. Running agents will stop
4. Status changes to CANCELLED
Re-running a Workflow
1. Click "Re-run" button
2. New execution starts with same workflow configuration
3. Navigates to new run page automatically
4. Previous run remains accessible
Editing During Execution
- Click "Edit Workflow" to modify the workflow definition
- Changes affect future runs, not the current execution
- Current run continues with original configuration
Loop Handling
Loop Cards
Loops appear as special cards showing:
- Loop label/name
- Number of agents inside
- Status avatars of agents within the loop
Loop Details
Clicking a loop card shows:
- Iteration breakdown
- Each cycle's agents and their statuses
- Combined thinking process logs
- Total cost across all iterations
Loop Status
- Loop card status reflects the overall loop state
- Individual agent statuses shown within loop details
- Iterations numbered sequentially
Best Practices
Monitoring Execution
1. Watch the Kanban Board: Cards move left to right as workflow progresses
2. Check Status Colors: Quick visual check of overall progress
3. Monitor Feedback Column: Respond promptly to agents needing input
4. Review Error Column: Check failed agents for issues
Providing Feedback
1. Review Output Carefully: Check agent response and generated files
2. Use Accept When Satisfied: Speeds up workflow completion
3. Provide Clear Feedback: Specific instructions help agents revise effectively
4. Check Generated Files: Preview media and documents before accepting
Troubleshooting
1. Check Error Cards: Click failed agents to see error details
2. Review Thinking Process: Understand why agent made decisions
3. Check Dependencies: Ensure parent agents completed successfully
4. Review Agent Instructions: Verify task clarity
Tips
- Keep Modals Open: They update automatically during execution
- Use Feedback Promptly: Agents wait for your input before continuing
- Review Files: Generated files may contain important outputs
- Monitor Costs: Check credit usage in agent details
- Check Execution Time: Helps optimize workflow performance
Summary
The Workflow Run Execution page is your command center for monitoring workflow execution. The Kanban board provides an intuitive visual representation of progress, while detailed modals give you complete insight into each agent's execution. Use the feedback system to guide agents when needed, and leverage the real-time updates to stay informed throughout the execution process.