ORM Code Generator
Overview
Generate type-safe ORM model classes, migration files, and repository patterns from existing database schemas or domain specifications. Supports Prisma, TypeORM, Sequelize, SQLAlchemy, Django ORM, and Drizzle ORM.
Prerequisites
- Database connection string or credentials for schema introspection
psql or mysql CLI for querying information_schema- Target ORM framework already installed in the project (
prisma, typeorm, sqlalchemy, etc.)
- Node.js/Python/Go runtime matching the target ORM
- Existing project structure to place generated models in the correct directory
Instructions
Introspect the database schema by querying information_schema.COLUMNS, information_schema.TABLE_CONSTRAINTS, and information_schema.KEY_COLUMN_USAGE to extract all tables, columns, data types, nullable flags, defaults, primary keys, foreign keys, and unique constraints.
For PostgreSQL, additionally query pg_catalog.pg_type for custom enum types and pg_catalog.pg_index for index definitions. For MySQL, query information_schema.STATISTICS for index details.
Map database column types to ORM field types:
varchar/text -> String / @Column('text')integer/bigint -> Int / @Column('int')boolean -> Boolean / @Column('boolean')timestamp/datetime -> DateTime / @Column('timestamp')jsonb/json -> Json / @Column('jsonb')uuid -> String with @default(uuid()) or uuid.uuid4- Custom enums -> Generate enum type definitions
Generate model classes with proper decorators/attributes:
- For Prisma: Generate
schema.prisma with model blocks, @id, @unique, @relation, and @default directives.
- For TypeORM: Generate entity classes with
@Entity(), @Column(), @PrimaryGeneratedColumn(), @ManyToOne(), @OneToMany() decorators.
- For SQLAlchemy: Generate model classes extending
Base with Column(), ForeignKey(), relationship(), and __tablename__.
- For Drizzle: Generate table definitions with
pgTable(), serial(), varchar(), timestamp(), and relations().
Generate relationship mappings from foreign key constraints. Detect one-to-one (unique FK), one-to-many, and many-to-many (junction table with two FKs) patterns automatically. Add both sides of each relationship with proper cascade options.
Create migration files that capture the current schema state. For Prisma: npx prisma migrate dev --name init. For TypeORM: generate migration with typeorm migration:generate. For Alembic: alembic revision --autogenerate.
Generate repository/service layer with common CRUD operations: findById, findAll with pagination, create, update, delete, and relationship-aware queries (findWithRelations).
Add validation decorators or constraints matching database CHECK constraints and NOT NULL columns. Use class-validator for TypeORM, Pydantic validators for SQLAlchemy, or Zod schemas for Prisma.
Generate TypeScript/Python type definitions or interfaces for API layer consumption, ensuring the ORM models and API types stay synchronized.
Validate generated models by running a test migration against a temporary database or by comparing the generated schema against the live database schema with a diff tool.
Output
- Model/entity files with full type annotations, decorators, and relationship mappings
- Migration files capturing the initial schema state
- Enum type definitions for database enum columns
- Repository/service classes with typed CRUD operations
- Validation schemas (Zod, class-validator, Pydantic) matching database constraints
- Type definition files for API layer consumption
Error Handling
| Error |
Cause |
Solution |
| Circular relationship dependency |
Two entities reference each other, causing import cycles |
Use lazy loading (() => RelatedEntity) in TypeORM; use ForwardRef in SQLAlchemy; split into separate files with deferred imports |
| Unknown column type mapping |
Database uses custom types, extensions, or domain types not in the standard mapping |
Add custom type mapping in generator config; use @Column({ type: 'text' }) as fallback; register custom transformers |
| Migration conflicts with existing data |
Generated migration adds NOT NULL columns without defaults |
Add default values to new columns; create a two-phase migration (add nullable, backfill, set NOT NULL) |
| Junction table not detected as many-to-many |
Junction table has extra columns beyond the two foreign keys |
Model as an explicit entity with two ManyToOne relationships instead of an implicit ManyToMany |
| Schema drift between ORM models and database |
Manual database changes not reflected in ORM code |
Run introspection again; use prisma db pull or sqlacodegen to regenerate; diff against existing models |
Examples
Prisma schema from PostgreSQL e-commerce database: Introspect 15 tables including users, orders, products, and categories. Generate schema.prisma with proper @relation directives, enum types for order status, and @default(autoincrement()) for serial columns. Output includes Zod validation schemas for each model.
TypeORM entities from MySQL SaaS application: Generate entity classes for a multi-tenant application with tenant isolation. Each entity includes a tenantId column with a custom @TenantAware decorator. Repository layer includes tenant-scoped query methods.
SQLAlchemy models from legacy database with naming conventions: Introspect a database with inconsistent naming (mix of camelCase and snake_case). Generate models with __tablename__ preserving original names while using Pythonic property names. Alembic migration captures the full schema.
Resources
Converted and distributed by TomeVault — claim your Tome and manage your conversions.
1---2name: jeremylongshore-claude-code-plugins-plus-skills-generating-o3description: ORM Code Generator4---5# ORM Code Generator67## Overview89Generate type-safe ORM model classes, migration files, and repository patterns from existing database schemas or domain specifications. Supports Prisma, TypeORM, Sequelize, SQLAlchemy, Django ORM, and Drizzle ORM.1011## Prerequisites1213- Database connection string or credentials for schema introspection14- `psql` or `mysql` CLI for querying `information_schema`15- Target ORM framework already installed in the project (`prisma`, `typeorm`, `sqlalchemy`, etc.)16- Node.js/Python/Go runtime matching the target ORM17- Existing project structure to place generated models in the correct directory1819## Instructions20211. Introspect the database schema by querying `information_schema.COLUMNS`, `information_schema.TABLE_CONSTRAINTS`, and `information_schema.KEY_COLUMN_USAGE` to extract all tables, columns, data types, nullable flags, defaults, primary keys, foreign keys, and unique constraints.22232. For PostgreSQL, additionally query `pg_catalog.pg_type` for custom enum types and `pg_catalog.pg_index` for index definitions. For MySQL, query `information_schema.STATISTICS` for index details.24253. Map database column types to ORM field types:26 - `varchar/text` -> `String` / `@Column('text')`27 - `integer/bigint` -> `Int` / `@Column('int')`28 - `boolean` -> `Boolean` / `@Column('boolean')`29 - `timestamp/datetime` -> `DateTime` / `@Column('timestamp')`30 - `jsonb/json` -> `Json` / `@Column('jsonb')`31 - `uuid` -> `String` with `@default(uuid())` or `uuid.uuid4`32 - Custom enums -> Generate enum type definitions33344. Generate model classes with proper decorators/attributes:35 - For **Prisma**: Generate `schema.prisma` with `model` blocks, `@id`, `@unique`, `@relation`, and `@default` directives.36 - For **TypeORM**: Generate entity classes with `@Entity()`, `@Column()`, `@PrimaryGeneratedColumn()`, `@ManyToOne()`, `@OneToMany()` decorators.37 - For **SQLAlchemy**: Generate model classes extending `Base` with `Column()`, `ForeignKey()`, `relationship()`, and `__tablename__`.38 - For **Drizzle**: Generate table definitions with `pgTable()`, `serial()`, `varchar()`, `timestamp()`, and `relations()`.39405. Generate relationship mappings from foreign key constraints. Detect one-to-one (unique FK), one-to-many, and many-to-many (junction table with two FKs) patterns automatically. Add both sides of each relationship with proper cascade options.41426. Create migration files that capture the current schema state. For Prisma: `npx prisma migrate dev --name init`. For TypeORM: generate migration with `typeorm migration:generate`. For Alembic: `alembic revision --autogenerate`.43447. Generate repository/service layer with common CRUD operations: `findById`, `findAll` with pagination, `create`, `update`, `delete`, and relationship-aware queries (`findWithRelations`).45468. Add validation decorators or constraints matching database CHECK constraints and NOT NULL columns. Use `class-validator` for TypeORM, Pydantic validators for SQLAlchemy, or Zod schemas for Prisma.47489. Generate TypeScript/Python type definitions or interfaces for API layer consumption, ensuring the ORM models and API types stay synchronized.495010. Validate generated models by running a test migration against a temporary database or by comparing the generated schema against the live database schema with a diff tool.5152## Output5354- **Model/entity files** with full type annotations, decorators, and relationship mappings55- **Migration files** capturing the initial schema state56- **Enum type definitions** for database enum columns57- **Repository/service classes** with typed CRUD operations58- **Validation schemas** (Zod, class-validator, Pydantic) matching database constraints59- **Type definition files** for API layer consumption6061## Error Handling6263| Error | Cause | Solution |64|-------|-------|---------|65| Circular relationship dependency | Two entities reference each other, causing import cycles | Use lazy loading (`() => RelatedEntity`) in TypeORM; use `ForwardRef` in SQLAlchemy; split into separate files with deferred imports |66| Unknown column type mapping | Database uses custom types, extensions, or domain types not in the standard mapping | Add custom type mapping in generator config; use `@Column({ type: 'text' })` as fallback; register custom transformers |67| Migration conflicts with existing data | Generated migration adds NOT NULL columns without defaults | Add default values to new columns; create a two-phase migration (add nullable, backfill, set NOT NULL) |68| Junction table not detected as many-to-many | Junction table has extra columns beyond the two foreign keys | Model as an explicit entity with two ManyToOne relationships instead of an implicit ManyToMany |69| Schema drift between ORM models and database | Manual database changes not reflected in ORM code | Run introspection again; use `prisma db pull` or `sqlacodegen` to regenerate; diff against existing models |7071## Examples7273**Prisma schema from PostgreSQL e-commerce database**: Introspect 15 tables including users, orders, products, and categories. Generate `schema.prisma` with proper `@relation` directives, enum types for order status, and `@default(autoincrement())` for serial columns. Output includes Zod validation schemas for each model.7475**TypeORM entities from MySQL SaaS application**: Generate entity classes for a multi-tenant application with tenant isolation. Each entity includes a `tenantId` column with a custom `@TenantAware` decorator. Repository layer includes tenant-scoped query methods.7677**SQLAlchemy models from legacy database with naming conventions**: Introspect a database with inconsistent naming (mix of camelCase and snake_case). Generate models with `__tablename__` preserving original names while using Pythonic property names. Alembic migration captures the full schema.7879## Resources8081- Prisma introspection: https://www.prisma.io/docs/orm/prisma-schema/introspection82- TypeORM entity documentation: https://typeorm.io/entities83- SQLAlchemy ORM tutorial: https://docs.sqlalchemy.org/en/20/orm/84- Drizzle ORM schema: https://orm.drizzle.team/docs/sql-schema-declaration85- Django inspectdb command: https://docs.djangoproject.com/en/5.0/howto/legacy-databases/8687---88> Converted and distributed by [TomeVault](https://tomevault.io/claim/jeremylongshore) — claim your Tome and manage your conversions.89<!-- tomevault:4.0:skill_md:2026-04-11 -->
tomevault-io