Troubleshooting

This guide provides a systematic approach to diagnosing and resolving issues within Codewords workflows. Follow these steps to identify the root cause of a problem, whether it's a clear error or unexpected behaviour.


1. The First Step: Ask the Codewords AI

Before diving into manual checks, always leverage the platform's built-in diagnostic capabilities. Codewords AI can analyse the last execution, interpret error messages, and often provide a direct explanation and solution.

How to Ask: Open the Codewords chat and use a prompt like the one below, replacing the placeholder with the specific URL of your workflow.

"Check the last execution of the workflow with the URL: [Your Workflow URL Here]. Why did it fail and how can I fix it?"


2. Investigating Unexpected Outputs

Sometimes a workflow completes successfully but produces strange or incorrect results. This requires a different approach.

Diagnostic Checklist:

  • Validate the Input Data: The principle of "Garbage In, Garbage Out" applies. Was the initial data that triggered the workflow correct and in the expected format?

  • Trace the Data Flow: Use the execution logs to follow the data from step to step. Where did it get modified unexpectedly? Was a step skipped due to incorrect logic (e.g., a filter)?

  • Isolate with Simple Data: Rerun the workflow with a very simple, predictable trigger. For example, use "Test Name" instead of a complex variable. If the simple test works, the issue lies in the complexity of your real data.

  • Review Your Logic: Double-check any filters, conditional paths, or data transformation steps. Is a filter too restrictive? Is a variable being mapped to the wrong field?

  • Consult the Examples: Review the documentation or examples for the specific service you're connecting to. Are you using the API or action correctly?


3. Escalation Procedure

If you have followed the steps above and cannot resolve the issue, it's time to escalate. To ensure a fast resolution, please provide the following information in your report.

Escalation Channels:

  • Primary: Post in the #dev-support Slack channel.

Information to Include:

  1. Workflow URL: A direct link to the workflow experiencing the issue.

  2. Execution ID: The specific ID of the failed or incorrect run.

  3. Problem Summary: A brief, one-sentence description of the issue.

  4. Expected vs. Actual Behaviour: Clearly state what you expected to happen and what actually happened.

  5. Troubleshooting Steps Taken: Briefly list the steps you've already tried from this guide. This prevents us from repeating work.

Last updated