ETL metadata injection driven by a json configuration file

I'm working on a set of jobs to drive metadata injection transformations via json configuration / database connections - trying to reduce the clutter of creating <x> jobs for <x> etl tasks. (Yay metadata!)

I've completed a POC proving I can drive various metadata injection steps through json conf files - with the exception of the transformation template - (which makes sense since the ui needs to be able to read the underlying transformation to refresh the available metadata inputs at design time).

I'm wondering what the best approach to this would be; ideally, I'd like to be able to replicate the same call that the metadata injection step is using (since it's probably only preventing the variable transformation template for the ui) - I'd like to perhaps replicate the call in javascript so I that it's a bit more visible to our developers (vs Java code). I'm wondering if this is a good approach (or not), and if not, where in the source I should be taking a peek - as well as if I should just be writing a piece of Java code to drive all this. As a note, I'm using Kettle 4.4.

There doesn't seem to be a lot of posts on the interwebs about this, so I thought I'd ask here first. Thanks for any help anyone can provide! :)