Per-entity overrides for the Prisma adapter.
Used to customise how the adapter addresses compound-key fields on a Prisma
model — in particular the PrismaCompositeKeyOptions.compositeKeyName
override that is mandatory when a model uses a named @@id constraint.
compositeKeyName: string
Override for the derived compound-key field name.
Prisma generates an unnamed @@id([a, b]) as a_b; a named
@@id([...], name: "custom") generates only custom. When the derived
name does not match the schema (e.g. a named @@id), set this to the
Prisma-side generated field name so findById/update/delete emit the
correct compound-key where argument.
keyColumns: readonly string[]
The primary-key columns for this entity, in Prisma schema declaration order.
Required when compositeKeyName is set: the adapter needs the
column names to build the compound-key object that Prisma expects inside
the where argument. Omitting it while naming a compositeKeyName leaves
the columns at the scalar default ['id'], and every composite key is
refused for missing the id column — measured against live PostgreSQL in
the M79 live suite. Supply every column, in declaration order, for both
unnamed @@id (derived field tenantId_userId) and named @@id models.
Defaults to ['id'] when absent (scalar-key path).