This article on Medium addresses the scopes of the domain model in hexagonal architecture.
I have sketched the following generic hexagonal architecture:
On the one hand, the secondary/output adapter receives the domain model ("model" for short) and maps it to a DAO for persistence.
On the other hand, the primary/input adapter receives only the "ExposedModel" interface. This is then mapped to a DTO for the client. Why is this intermediate step needed? The output adapter also receives the complete model.