Seed data strategy

Why it matters

Ad-hoc seed scripts rot. Every engineer needs a known-good dataset that matches the current schema. Meridian seeds are idempotent, ordered, and pinned to a migration version so you never waste time debugging stale fixtures.

Seed file layout

prisma/
├── seeds/
│   ├── 001_tenants.ts
│   ├── 002_users.ts
│   ├── 003_licenses.ts
│   └── runner.ts

Each file exports a seed() function that receives the Prisma client. The runner executes them in numeric order inside a single transaction.

Idempotency

Every seed step checks existence before inserting. Use upsert or a findFirst guard so npx prisma db seed is safe to run repeatedly.

CI integration

The seed runner is invoked after prisma migrate deploy in GitHub Actions. A smoke-test step then validates key rows — tenant count, admin user existence, and active license count — so broken seeds fail the pipeline before they reach staging.

Resetting

Local reset is a single command:

npx prisma migrate reset --force

This drops the database, re-runs all migrations, and executes the seed suite — giving you a pristine environment in under 10 seconds.