Requesting Response Field Indicating If API Returns a "can not do that" text

API Feedback
1

We are taking the content.text.value of a message response and adding material before and after the value. This works great if the API returns the results we expect.

There are times though where the content.text.value may be “Sorry I am not allowed to do that. Please contact the site administrators for assistance.” or “I’m sorry, but I can’t fulfill this request.” or some varient on this where the API is basically declines from generating an expected response. I will call this a “decline to respond”

In the cases where the API issues a “decline to respond” we don’t want to prepend or append the content.text.value - just want to leave it as is.

Currently, there is not an easy way to determine that the API has provided a “decline to respond” (as far as I can see). And doing a read of the content.text.value for a start of “Sorry” or “I’m sorry” or “I can’t” doesn’t create a beleif in consistency (maybe the expected response starts with that, or over time or AI models, the beginning of a “decline to respond” starts using different language).

So, we can not assume that even our parsing the content.text.value can be trusted to be accurate over time, or right now.

Can you please add an additional field that flags when the API returns a “decline to respond” response? That would be super super helpful!

Thank you

2

I don’t know if I understand your set-up in full but technically you could work out a solution whereby you have the API return a JSON object that includes both the actual API response and a standardized classifier that is indicative of the nature of response (could be binary value or some such). You can then just parse the JSON and based on the value of the classifier you proceed in the desired way.