I'm a bit confused. Should we write ourselfs the resolvers with prisma? Like in this boilerplate: https://github.com/graphql-boilerplates/node-graphql-server/blob/master/basic/src/index.js

You could also use

forwardTo

if you wanna use the default resolvers

From my understanding, prisma is basically just a smart database service that exposes CRUD functionality. When writing a graphql server, you can/should put some middlewhere between your client and the prisma service that implements business logic. In the example you referenced, that is implemented in node.