For a long time I've thought that the idea of having an entire, Turing-complete layer between the database and the HTTP API felt like overkill. You know your collections, you know the shape of your data, auth doesn't vary much between orgs. Unless you're doing interesting computations after querying, why do you have to <i>program</i> those endpoints over and over again for each entity in each project? It just seems silly.<p>I'm honestly surprised there aren't more projects like this