d1-drizzle-schema

d1-drizzle-schema generates Drizzle ORM schemas for Cloudflare D1 databases using D1-correct patterns. It solves a specific class of subtle bugs: D1 is SQLite-based but differs from standard SQLite in ways that trip up default patterns — foreign keys are always enforced and cannot be disabled, there are no native BOOLEAN or DATETIME types, the bound-parameter limit is 100 (not ~999), JSON support is always available, and writes are single-threaded. Using ordinary SQLite conventions against D1 produces code that looks right but misbehaves.

The skill walks a seven-step workflow: describe the data model, generate the Drizzle schema with D1-correct column patterns (text UUID primary keys via `crypto.randomUUID()`, booleans as `integer({ mode: 'boolean' })`, timestamps as `integer({ mode: 'timestamp' })`, enums stored as TEXT, typed JSON as TEXT, and always-enforced foreign key references), add Drizzle relations, export inferred types, set up `drizzle.config.ts`, add wrangler migration scripts, and generate a DATABASE_SCHEMA.md for future sessions. It includes a bulk-insert batching pattern that respects the 100-parameter limit and D1 runtime query examples. Supporting reference files cover the full column-type catalog and D1-versus-SQLite specifics, and template assets provide a starter config and example schema. The skill is marked claude-code-only.

Target users are developers building on Cloudflare Workers with D1 who use Drizzle ORM and want schemas that avoid D1's footguns from the start. Typical use cases include scaffolding a new D1 data layer, adding tables to an existing project, and documenting a schema. The migration commands it suggests use standard Cloudflare tooling (`drizzle-kit generate`, `wrangler d1 migrations apply`) and are ordinary, non-destructive developer operations.