I like the "single source of truth" concept. It makes a lot of sense IMO.
Does it generate the handlers for you or do you have to write them by hand and hook to the Swagger definition?
The server provides dependency injection so that when you implement your handler-level function it can access several re-usable services to do the "real work". Generally my handlers are limited to getting the parameters off the request, passing them on to a (chain of) service(s) and getting the response back out to the client.
I hope the wiki has some good additional info if you're interested: https://github.com/blueoakjs/blueoak-server/wiki
How much does Ramses integrate with the data source "automatically"? Is it like StrongLoop/LoopBack?
We've found that for most of our projects we're not connecting directly to a database, but instead trying to present an (mobile or web) app-specific API that abstracts/unifies/focusses the many data sources and services that may be needed.
For example, one recent project involved data coming from some old SOAP services, new HTTP (but not exactly RESTful) endpoints, and several public APIs too.