I have been able to use YAML to import data structures. When I did it, it was not because I was trying to connect to a service, but because I had 250 properties defined on my case (from ideation) and I wanted to quickly reorganize those properties into classes. When I did it, I used a spreadsheet to work through what my target data model should be and then used copilot to generate a YAML to import the target data model into my blueprint so I could import it. (Co-pilot was not perfect.)
Here is an image (can’t attach a file) of what your YAML should look like to be accepted:
You don’t have to include a description, but if you do, the properties will have that description.
if you get a data type wrong, for example using Text instead of string, the property may not import. I have also had cases with large YAMLs where I had to search for errors (thanks co-pilot) to get them to import.
If you see an object identified after doing your upload, you are half way there.
Your schema will be tested more thoroughly when hitting submit and other errors can prevent the import. This can be very aggravating.
Post upload, I recommend updating your labels for your fields as the labels and IDs will match what you imported. Doing this, you should be able to cleanly import your defined classes.
I also recommend doing a test import into another blueprint to verify what you are getting as you can’t delete embedded data classes should you decide you want to reimport after making corrections.