What problem does it solve? Applying schema changes during prototyping or development often requires writing migration files even when you just want to sync a Drizzle schema to a live database quickly. This Skill guides the correct use of drizzle-kit push and the push(...) SDK so schema diffs are executed directly against the database with structured, machine-readable results. ## Core Features & Use Cases - Direct schema sync: Push a Drizzle schema to a live PostgreSQL, MySQL, SQLite, MSSQL, CockroachDB, SingleStore, or Turso database without generating migration files. - Structured JSON envelope: Decode the discriminated-union response (ok, no_changes, missing_hints, error) from both the CLI (--output json) and the SDK, with matching exit codes. - Destructive-change safety: Detect confirm_data_loss situations (dropping non-empty tables or columns, SQLite table recreation) and resolve them through the hints loop before re-invoking. - Use Case: While prototyping a new feature, you add a users.legacy_id column and run drizzle-kit push --output json; the envelope reports missing_hints for a destructive change, you supply the confirmation hints, and the schema is applied. ## Quick Start Ask the AI to push your Drizzle schema to the development database using drizzle-kit push with JSON output and interpret the response envelope.