database-design

name	database-design
description	Covers schema design, migrations, query optimization, indexing strategies, and normalization tradeoffs. Use when designing tables, writing migrations, optimizing queries, or planning index strategies. Provides patterns for database administration and development. Not for ORM usage in application code (use language-specific development skills) or infrastructure orchestration (use infrastructure-management).
user-invocable	false
allowed-tools	Read, Write, Edit, Glob, Grep, Bash(psql:*, sqlite3:*, mysql:*)
version	0.3.1

Database Skill

Domain knowledge for database administration and development. Covers schema design, migrations, query optimization, and indexing strategies.

Critical Rules

• Data integrity first -- constraints catch bugs that code misses; always define foreign keys, NOT NULLs, and check constraints at the database level
• Backward compatibility -- migrations must not break running code; use safe online operations (ADD COLUMN nullable, ADD INDEX CONCURRENTLY) over unsafe ones
• Measure before optimizing -- use EXPLAIN ANALYZE, not intuition; never add indexes or denormalize without evidence of a performance problem
• Plan for scale -- design decisions are expensive to change; consider write patterns, cardinality, and growth from the start

Verification

• New tables have appropriate primary keys, foreign keys, and NOT NULL constraints
• Migrations use only safe online operations, or document a maintenance window if unsafe operations are required
• Query changes are validated with EXPLAIN ANALYZE showing expected plan improvements

Quick Reference

Primary Key Selection

Type	Use When	Avoid When
UUID v4	Distributed systems, no sequential exposure	Need sortability, space-constrained
ULID	Need sortability + uniqueness	Legacy system compatibility
Serial/Identity	Single database, simple use case	Distributed writes, ID exposure concerns

Migration Safety

Safe (online):     ADD COLUMN (nullable), ADD INDEX CONCURRENTLY, CREATE TABLE
Unsafe (offline):  ADD COLUMN NOT NULL (without default), DROP COLUMN, RENAME

Index Decision Tree

High cardinality + equality lookups  → B-tree (default)
Low cardinality + many values        → Consider partial index
Array/JSON containment               → GIN
Geometric/range queries              → GiST
Exact equality only                  → Hash (rare)

Topics

Topic	Reference	Use When
Schema Design	schema-design.md	Creating tables, choosing keys, audit patterns
Migrations	migrations.md	Writing safe, reversible migrations
Query Optimization	query-optimization.md	Debugging slow queries, N+1 detection
Indexing	indexing.md	Choosing index types, composite index order

When to Use This Skill

• Designing new tables or modifying existing schemas
• Writing or reviewing database migrations
• Optimizing slow queries
• Planning index strategies
• Evaluating normalization vs denormalization tradeoffs

Related Skills

• See foundations for code style, TDD, and verification principles
• See infrastructure-management for connection pooling and deployment patterns