Hello Community,
I would like to ask about the expected and recommended way to trace HTTP errors (especially 404) when using Virtual Agent + IntegrationHub (RESTMessageV2).
Background
I have a Virtual Agent (VA) that triggers a parent Flow, which in turn calls a child Subflow.
Inside the child Subflow, I use an IntegrationHub action based on RESTMessageV2 to call the ServiceNow Table API.
For testing purposes, I intentionally configured the action so that it returns HTTP 404:
- In the IntegrationHub action (EXT API – Incident CRUD via RESTMessageV2)
- I set target_id to a non-existent sys_id
- When running the Action test, I clearly see: Error: HTTP 404 - No Record found
So the 404 is definitely happening at the REST level.
The Problem
However, when the same action is executed via Virtual Agent conversation:
- The VA conversation completes without a visible error
- The Flow Error Handler is not triggered
- API Audit Log / custom audit tables only show 20x statuses
- There is no built-in log where I can later confirm that a 404 occurred
- Flow execution settings / execution details also do not show the 404
From my investigation, this seems to be caused by the IntegrationHub design:
2xx / 404 → verified non-2xx → throws
So 404 is treated as a verified result, not an exception.
Why this is concerning
From an operational and audit perspective:
- A real error did occur at the HTTP layer
- But Virtual Agent executions provide no traceable evidence of it
- After the conversation finishes, there is no way to prove or review that the API returned 404
- This makes troubleshooting and post-incident analysis very difficult
My Questions
- Is it by design that Virtual Agent executions do not retain or expose HTTP error details such as 404?
- Is there any standard or recommended configuration to make such errors traceable (without turning 404 into a thrown exception)?
- Is the only supported approach to explicitly log status, method, etc. via a custom Create Record step in the Flow/Subflow?
- Are there any best practices for error visibility when using:
- Virtual Agent
- Flow Designer
- IntegrationHub (RESTMessageV2)
What I am trying to confirm
I am not looking for how to force an error.
I want to understand the correct architectural approach to ensure that:
“If an HTTP error occurs during a VA-driven integration, it can be reviewed later.”
Any guidance from ServiceNow staff or experienced architects would be greatly appreciated.
Thank you.
Koyama
#workflow #flow&actions
