The project has been started around 2016. Initially, it didn't get very far, because of being overambitious. In particular, the code styles have been difficult. Nevertheless, it has been restarted two, or three times, before being put to actual use in one of my employers projects. (Thanks to Holger Döppert for tasking me with that.)
As of this writing, the project is usable, and most of the intended features are present: We've got an API for working with annotations, we can deal with generics, the two important code styles are available, and the problems in building a class hierarchy of builders, have been resolved.
On the other hand, I am not satisfied with the class structure. (For example, the classes Block, and CodeBlock are definitely somewhat redundant, and I am still not completely happy with the builders. (For example, within a block, I'd like to see the ability to return to the outer block.)
That said, the general ideas have been borrowed from a mature predecessor (JaxMeJS), and the well known, and proven concepts have been adopted, and even refined. There is no reason to assume that these will need to be changed greatly.
In general, the framework should be ready for inspection, for proof of concept work, smaller projects, and the like. It might be a risk to use it for large-scale applications.
However, if you are ready to adjust to a few required changes over time upon adopting a new version should be adopted, then Jsgen should be able to satisfy your expectations quite decently.
Thus, as of 2019, I'd say: Give it a try. It should be worth it.
