ixtab:
In response to your questions on the calibre plugin Kindle Collections I'd recommend downloading or installing the plugin file, and then just unzip the python code to look at the details. Its been a long time since I looked at this but I'll try to give some answers. Since I don't have a Kindle touch I can't do any testing for you, but I think handling the format should be straightforward given what you've already done.
The json format you gave looks correct:
[QUOTE=ixtab;2167095]
Code:
{"Sample Series@en-US": {"items": ["#7af2ad8e-f04e-4797-bab4-8dfc0277210d^EBOK", "#B004UBEKXI^EBOK"], "lastAccess": 1234567890000}}
[/code]
"items" must always be there, but can be an empty array []
"lastAccess" must always be there, but can be 0, or whatever date you want to set
Those are the only fields that I can recall.
The CDE type may not matter for your software if all you are doing is importing the json file created by the calibre plugin. There are types like EBOK, MAGZ for magazine, PDOC for personal document, etc. The string is just the book code that the Kindle uses to identify your book. The shorter ones starting with # are Amazon books, the longer starting with * are hash numbers for side-loaded books. I don't know if the algorithmn for the book codes is the same on the Touch.