← ClaudeAtlas

postgres-drizzlelisted

Engineer plain PostgreSQL databases with Drizzle: schemas, constraints, migrations, expand-contract rollouts, transactions, isolation, locking, tenant boundaries and RLS, indexes, query plans, pagination, data types, pooling, and generated SQL review. Use for PostgreSQL/Drizzle schema changes, migration safety, query performance, concurrency bugs, or database invariants. Retrieval-first for version-sensitive Drizzle APIs and commands. This skill does not own Supabase services, generic API design, or application authentication.
ongkipro/dotfiles · ★ 0 · API & Backend · score 53
Install: claude install-skill ongkipro/dotfiles
# PostgreSQL + Drizzle Engineering Make each database invariant explicit and enforce it at the lowest correct layer. A TypeScript type is not a database guarantee; every writer must meet the same constraints. Treat generated SQL, migrations, connection configuration, and RLS policies as privileged code. ## Trigger and scope Use this skill for plain PostgreSQL and Drizzle schema/query engineering, including: - tables, keys, constraints, relationships, data types, defaults, and generated values; - migration generation/review, expand-contract rollout, backfills, and drift diagnosis; - transaction boundaries, isolation anomalies, locks, retries, and concurrent writes; - tenant keys, plain PostgreSQL RLS, indexes, query plans, and pagination; - driver/pool/serverless connection behavior and database-level verification. Do not use it as the owner of Supabase Auth, Storage, Realtime, Edge Functions, its platform CLI lifecycle, or `auth.uid()` conventions; use `supabase-stack`. Do not redesign HTTP/GraphQL contracts or implement authentication here. ## Inspect first Before proposing or running a command: 1. Read repository instructions, package scripts, lockfile, installed Drizzle packages, `drizzle.config.*`, schema exports, migration journal/snapshots, and all generated SQL. 2. Trace every affected query and writer, including jobs, admin tools, imports, auth adapters, seeds, tests, and direct SQL. Identify who owns the transaction. 3. Identify the PostgreSQL server ve