Skip to main content

North star

Make production-grade database schema deployment as ergonomic as writing a JSX component — with a Time Machine for your schema and AI review built in.

That's the whole pitch.

The five capabilities

A miss on any one isn't fatal; together they are the wedge.

CapabilityDrizzle / FlywayPrismaDB-X
Declarative schemaDSLPSLJSX
Time Machine for schemasnapshot + restore + history + diff
Plan / preview before applyshadow-DB + lock report
AI-reviewable diffMCP server
Schema → ORM typesmanualbuilt-in (for Prisma)export for any ORM

Audience

  • Primary: application engineering teams that already use Drizzle / Prisma / Kysely and currently deploy schema changes by hand, drizzle-kit push, or Flyway. They want a CI-safe gate with rollback.
  • Secondary: Infra-X users who want richer DB tooling than the reference @infra-x/postgres-library and prefer the focused db-x CLI.
  • Not: greenfield solo devs running migrations from their laptop — Drizzle is great for them; we don't displace it.

Non-goals

  • No ORM. No query DSL, no runtime data-access helpers, no connection pool. We do not compete with Drizzle's query layer or Prisma Client.
  • No data migrations as a first-class concept. Schema only. Backfills happen via explicit <SeedData> or your ORM's scripts; we don't model them.
  • No cross-database schema diffs. v0.x targets one database per apply tree.
  • No new IaC pitch. DB-X is not "Terraform for databases" — it's a focused tool on top of an existing runtime.

How we'll know it worked

  • A team that uses Drizzle in production replaces drizzle-kit push with db-x apply in their CI pipeline.
  • A PR-review tool (Claude / Cursor) answers "is this schema change safe to merge?" by talking to db-x mcp against the real DB.
  • An on-call engineer rolls back a bad schema deploy with one command in under 60 seconds.