I looked into this and you're using a homebrew class as you said, which isn't supported at the moment because of the way the converter matches classes. The actual class data can't be read from the PDF unfortunately, so the application uses class data files based on the name of the class. It does this for things like races, spells and items as well.
I do want to look if it's possible to extract the needed data from a JS file. If it has a standardized format, there should be a way. Ideally you could then put your JS class files in a folder so the converter is aware of them.
Can you provide me your JS file so I have a starting point? Thanks!