I started with the decision tree template and that worked fine. However once I started customizing the skill, it quit immediately after executing the launch request correctly. For some reason it can't seem to move into the recommendation intent. Instead it goes into SessionEnded either because of user_initiated (which I didn't) or exceeded max prompts. I have attached the JSON file.
There are some other things I noticed during my debugging attempts in the device log:
Differences between these logs starting from below (that’s where the processing starts):
- After expect speech the buggy skill has a render detail log, the functioning one does not
- After text.text message the buggy skill goes into settings.settingsupdated whereas the functioning skill returns speechsynthesizer.speech finished
Furthermore offsetInMilliseconds is 1000 ms for the working and 0 for the non-working skill. I wanted to set this manually but couldn't find where to do that.