From harness-claude
Manages Prisma database schema evolution using migrate dev/deploy/reset, baselining existing DBs, drift detection, history resolution, and production deployment.
npx claudepluginhub intense-visions/harness-engineering --plugin harness-claudeThis skill uses the workspace's default tool permissions.
> Manage database schema evolution with prisma migrate dev/deploy/reset and migration history
Evolves database schemas safely with Drizzle/Prisma migrations, zero-downtime strategies, rollbacks, backfills, and data validation for production without downtime or data loss.
Guides safe schema changes, data migrations, rollbacks, and zero-downtime deployments for PostgreSQL, MySQL, and ORMs like Prisma, Drizzle, Django.
Provides best practices for safe database migrations: schema changes, data backfills, rollbacks, zero-downtime deploys for PostgreSQL, MySQL, Prisma, Drizzle, Django.
Share bugs, ideas, or general feedback.
Manage database schema evolution with prisma migrate dev/deploy/reset and migration history
schema.prisma, run:npx prisma migrate dev --name describe_the_change
This generates a migration SQL file, applies it, and regenerates Prisma Client.
migrate dev in production. Use:npx prisma migrate deploy
This applies all pending migrations without generating new ones.
npx prisma migrate reset
mkdir -p prisma/migrations/0_init
npx prisma migrate diff --from-empty --to-schema-datamodel prisma/schema.prisma --script > prisma/migrations/0_init/migration.sql
npx prisma migrate resolve --applied 0_init
npx prisma migrate diff --from-schema-datamodel prisma/schema.prisma --to-schema-datasource prisma/schema.prisma
Customize generated SQL — after migrate dev creates a migration file, edit it before committing. Prisma will not overwrite your edits on subsequent runs.
Handle failed migrations — if a migration partially applied:
# Fix the database state manually, then mark as applied:
npx prisma migrate resolve --applied MIGRATION_NAME
# Or mark as rolled back:
npx prisma migrate resolve --rolled-back MIGRATION_NAME
migrate dev requires a shadow database to detect drift. Configure it explicitly if your provider does not allow automatic creation:datasource db {
provider = "postgresql"
url = env("DATABASE_URL")
shadowDatabaseUrl = env("SHADOW_DATABASE_URL")
}
Always commit migration files (prisma/migrations/) to version control. They are the source of truth for production.
Never delete or reorder migration directories once they have been applied to any environment.
Prisma Migrate uses a migrations table (_prisma_migrations) to track which migrations have been applied. Each migration is a directory containing a migration.sql file and is applied in lexicographic order.
Migration history vs schema state: The migration history is the authoritative record. The schema file describes the desired end state. If these diverge (drift), migrate dev will detect it via the shadow database.
Shadow database mechanics: During migrate dev, Prisma creates a temporary shadow database, applies all existing migrations to it, computes the diff against your schema, and generates the new migration. The shadow database is dropped afterward. This means you need CREATE DATABASE privileges in development.
Data migrations: Prisma generates only DDL (schema changes). If you need DML (data changes — backfills, column value transforms), add raw SQL to the generated migration file before committing.
Trade-offs:
migrate dev is safe and reversible (via reset), but slow on large schemas due to shadow database overheadmigrate deploy is fast but irreversible — there is no built-in rollback. Write reverse migrations manually if neededdb push skips migration history entirely — use it only for rapid prototyping, never in a shared environmentCommon mistakes:
migrate dev in CI — it generates files and requires a shadow database. Use migrate deploy insteadhttps://prisma.io/docs/orm/prisma-migrate