They aren't making something for the JS ecosystem: they're defining a new data serialization format built on explicitly defined input parsers (Scroll), and for one reason or another, its current implementation is in JS.

It turns out that creating a data serialization format built on explicitly defined parsers has uses beyond the ontological use cases, and there are several expressions of Scroll that are useful for technical use cases, such as Stamp for templating and ScrollSets[1] for structured data. The ScrollSets page is an excellent demonstration of how Scroll can be used to serialize data for human- and machine-consumption.

[1]: https://scroll.pub/blog/scrollsets.html