Roadmap

Relayer is in early development. The core query layer is stable and usable. Future packages will build on top of it to provide end-to-end CRUD automation.

Current packages

Package	Status	Description
@relayerjs/core	Published	ORM-agnostic types, contracts, and decorators
@relayerjs/drizzle	Published	Drizzle ORM adapter with full query DSL
@relayerjs/next	Published	Next.js App Router integration

Recent changes

Class-based entity model (v0.4)

Entity definitions moved from config objects to class-based models with decorators:

const UserEntity = createRelayerEntity(schema, 'users');

class User extends UserEntity {
  @UserEntity.computed({ resolve: ({ table, sql }) => sql`...` })
  fullName!: string;

  @UserEntity.derived({ query: ..., on: ... })
  postsCount!: number;
}

This replaces the previous { fields: { fullName: { type: FieldType.Computed, ... } } } config.

Planned packages

@relayerjs/rest

Auto-generate REST CRUD endpoints from your Relayer entities. Planned framework support: Express, Fastify, Hono.

The query DSL is already JSON-serializable, making it straightforward to accept where, select, orderBy as query parameters or request body fields.

@relayerjs/nest

NestJS module providing: CRUD controllers with automatic routing, GraphQL resolvers, decorator-based configuration.

@relayerjs/graphql

Standalone GraphQL schema generation from Relayer entities: auto-generated types, queries, and mutations, filter input types matching the Relayer DSL, relation loading via DataLoader pattern.

@relayerjs/react

React client library with hooks for querying Relayer endpoints: useQuery / useMutation hooks, type-safe integration with @relayerjs/rest or @relayerjs/next, optimistic updates and cache management.

Future ORM adapters

The long-term goal is a single unified query interface regardless of the underlying ORM. Potential future adapters:

• TypeORM
• Kysely
• MikroORM
• Prisma (as an alternative query layer)

Contributing

Contributions are welcome. If you are interested in any of the planned packages, open an issue on GitHub to discuss the approach before starting work.