We currently have the conversation.repetition.detected event that looks for repeating lines. It appears from the codeand my own local testing that we do not fire this event when the request was cancelled. I believe this is a major gap because true thought repetition within a request would probably lead to the user intervening and cancelling before it completes most of the time. We would like to use this metric to track occurrence of thought looping within a single request. We may also be able to use other tools in CAPI to detect this, but if it is a relatively easy change, I think it could be worth looking at this from the client side as well.

FYI @isidorn , @lramos15