Looks cool. But as you expose an endless number of queries you don’t know which indexes to support. Seems pretty easy to attack the server finding it’s slow non indexed queries and bringing your system to a hault. But even if you protect against that with persisted queries you’ll still need to define indexes for every query the client makes moving the work to the backend again. Seems like it may just be easier to allow the backend to control what is and isn’t queryable. Or am I missing something?