SeedForge

Schema-aware Prisma seed generator.

SeedForge reads schema.prisma, analyzes models and relations, and generates a working prisma/seed.ts. It is built for development workflows where schemas change often and seed scripts become expensive to maintain by hand.

Why

Manual Prisma seeds usually drift in the same places:

• required fields change
• relation order breaks
• new models are missed
• local environments stop matching each other

SeedForge generates deterministic relational seed code from the schema instead of relying on a hand-maintained seed file.

What it does

• Parses Prisma schema metadata with DMMF
• Builds a dependency graph from model relations
• Orders inserts to satisfy foreign keys
• Generates deterministic values for required scalars, enums, and scalar lists
• Supports model-level count, perParent, and disabled config
• Generates Prisma 7-compatible Postgres seeds using @prisma/adapter-pg
• Makes generated seeds idempotent by clearing existing data in reverse dependency order

Example dependency chain:

Category -> Product -> ProductVariant
Customer -> Order -> OrderItem -> Payment

Install

npm install -D seedforge

Quick start

1. Generate a seed:

npx seedforge generate

2. Generate the Prisma client:

npx prisma generate

3. Run the seed:

npx prisma db seed

Recommended workflow after schema changes:

npx prisma migrate dev
npx seedforge generate --force
npx prisma generate
npx prisma db seed

Prisma 7 setup

SeedForge supports Prisma 7-style config-driven datasource URLs.

Example prisma.config.ts:

import "dotenv/config";
import { defineConfig, env } from "prisma/config";

export default defineConfig({
  schema: "prisma/schema.prisma",
  migrations: {
    path: "prisma/migrations",
    seed: "tsx prisma/seed.ts",
  },
  datasource: {
    url: env("DATABASE_URL"),
  },
});

Example datasource block in prisma/schema.prisma:

datasource db {
  provider = "postgresql"
}

For Prisma 7 Postgres projects, generated seeds initialize Prisma Client with @prisma/adapter-pg.

Commands

seedforge generate

Generate prisma/seed.ts from the current schema.

Useful flags:

• --schema <path>: custom schema path
• --output <path>: custom output file
• --dry-run: print generated seed without writing
• --force: overwrite an existing file

seedforge init

Create seedforge.config.ts.

seedforge doctor

Validate schema loading and dependency ordering.

seedforge visualize

Print an ASCII dependency graph.

seedforge clean

Remove the generated seed file.

Configuration

Create seedforge.config.ts in the project root:

export default {
  global: {
    count: 5,
  },
  models: {
    Category: { count: 6 },
    Product: { perParent: 8 },
    ProductVariant: { perParent: 3 },
    Payment: { disabled: true },
  },
};

Rules:

• count: total records for a model
• perParent: records created per required parent relation
• disabled: skip a model entirely

Generated seed behavior

Generated seeds are deterministic for the same schema and config. That makes them easier to diff, rerun, and debug.

The generated file:

• imports Prisma Client
• clears existing records in reverse dependency order
• creates parent models before children
• reuses created record IDs for required foreign keys
• emits enum values from the schema
• emits scalar lists when required

Supported scope

Works best for development seeding on relational Prisma schemas, especially:

• ecommerce apps
• SaaS products
• dashboards and admin tools
• content systems

Databases:

• PostgreSQL
• MySQL
• SQLite
• CockroachDB
• MongoDB support is best-effort at schema parsing level

Limits

SeedForge generates valid scaffolding, not domain-perfect fixture data.

You may still want to customize:

• business-specific enums and statuses
• auth-sensitive fields
• highly constrained unique data
• special relation strategies for edge-case schemas

Local development

npm install
npm run build
npm run dev -- generate --dry-run

Examples

See examples/README.md for sample schemas and configs.

License

MIT