How do I troubleshoot a data or function node that returns unexpected data?

A Data or Function node can produce unexpected data because of a malformed input payload, incorrect filter conditions, missing or incorrect attributes, service-side issues, or problems with the underlying source data.

Use the following steps to determine whether the issue originates from the input payload, node configuration, connected service, or source data:

  1. Go to AI Agent Studio and open the Workflows workflows tab.
  2. Select Edit Edit to modify a workflow.
  3. From the toolbar, select Debug Debug.
  4. Enter a prompt in Ask Oracle and submit it to run the workflow.
  5. Inspect the Function node input to verify that it contains the expected payload.
  6. Confirm that all attribute names, values, and operators are correct and match the function or service requirements.
  7. Inspect the Function node output to determine whether the unexpected data originates from the function itself.
  8. If the input is incorrect, trace the workflow upstream and inspect the payload-generation node to identify where the data became invalid.
  9. If the input is correct but the output is incorrect, investigate the selected function, tool, API endpoint, service configuration, or the underlying source data for errors or inconsistencies.
By validating the input, output, and upstream data flow, you can isolate whether the issue originates in the workflow, the function, the connected service, or the source data.