Skip to content

Debugging Workflows

Learn how to identify and fix issues in your workflows.

Every workflow run is logged in the execution history.

  1. Open your workflow
  2. Click the History tab
  3. See all past executions with status

Click any execution to see:

  • Start time and duration
  • Trigger data that started the run
  • Each node’s input and output
  • Errors with stack traces
  • Final status and output
StateIconDescription
Running🔵Currently executing
CompletedFinished successfully
FailedEncountered an error
CancelledManually stopped
Waiting⏸️Paused (delay or waiting)

Cause: AI agent processing error

Debug steps:

  1. Check the task description for issues
  2. Verify input context exists and has values
  3. Check if inputs are too long
  4. Review the agent configuration

Solutions:

  • Simplify the task description
  • Check the agent’s instructions and tools
  • Break into multiple smaller nodes

Cause: Configuration issue

Debug steps:

  1. Verify the workflow is Published
  2. Check trigger configuration
  3. For scheduled workflows: check timezone settings
  4. Check the execution history for any runs

Cause: Previous node didn’t output expected data

Debug steps:

  1. Check if previous nodes completed successfully
  2. Verify the previous node’s output structure
  3. Check execution history for node outputs
  4. Look for null/empty values in context

Cause: Performance issues

Debug steps:

  1. Check execution history for timing per node
  2. Identify the slowest nodes
  3. Look for unnecessary sequential operations
  4. Check for large data transfers

Solutions:

  • Parallelize independent operations
  • Reduce data transferred between nodes
  • Cache repeated lookups
  • Optimize AI prompts to be more concise

View detailed execution information:

  1. Open execution history
  2. Click an execution
  3. View node execution details, inputs, and outputs

When nodes fail:

  • Errors are logged in the execution history
  • You can view detailed error messages and stack traces
  • Failed nodes are marked in the workflow execution view
  • Workflows may continue or stop depending on configuration
  1. Test with realistic data: Use production-like test data
  2. Test edge cases: Empty arrays, null values, very long strings
  3. Test error paths: Intentionally trigger errors
  4. Test timing: Run at different times if schedule-dependent
  5. Test with fresh state: Don’t rely on leftover test data

Track workflow execution:

  • Execution history: View all runs with status and duration
  • Success rate: Monitor successful vs failed runs
  • Performance: Track execution time per node
  • Error frequency: Identify patterns in failures

If you can’t resolve an issue:

  1. Check the FAQ
  2. Search the Troubleshooting guide
  3. Contact support with:
    • Workflow ID
    • Execution ID
    • Steps to reproduce
    • Expected vs actual behavior