# Go Database > Greenfield Go database workflow for database/sql, pgx, migrations, transactions, pooling, query correctness, test databases, and safe data boundaries. Use when this capability is needed. - Skill: `tomevault-io/go-database-2` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/go-database-2` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/go-database-2/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Data & Analytics - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/go-database-2 --- # Go Database ## Rule Make persistence explicit, transactional, context-aware, and isolated from real mutable systems. Keep SQL close enough to review and data access behind small boundaries. ## Hard Stops Stop before: - Touching production or developer databases without explicit approval and backup/safety plan. - Running destructive migrations, changing schema, or changing data retention semantics without rollback and compatibility discussion. - Adding an ORM, query generator, migration tool, or container dependency without approval. - Using real credentials or committing generated secrets. ## Defaults - Use `database/sql` for generic SQL access and `jackc/pgx/v5` for PostgreSQL driver/pool needs. - Prefer explicit SQL and small repository/store types over full ORMs in greenfield Go. - Use `golang-migrate/migrate/v4` for versioned SQL migrations when migration tooling is needed. - Pass contexts into queries and transactions. - Make transaction boundaries explicit; do not hide large workflows in auto-transactions. - Map database errors to domain errors at the data boundary. - Configure pool limits and timeouts intentionally. ## Testing Use isolated test databases, transaction rollbacks, temporary schemas, or approved `testcontainers-go` when real database behavior matters. Do not use production-like shared state by default. Test migrations up/down when the tool supports it, constraints, not-found behavior, uniqueness conflicts, and transaction rollback. ## Workflow 1. Define schema, migrations, queries, transaction boundaries, indexes, and failure behavior. 2. Choose stdlib `database/sql`, pgx, or project-approved tool based on needs. 3. Implement data access behind a small interface or concrete store consumed by services. 4. Add isolated tests and fixtures. 5. Run migrations/tests, `go test ./...`, race tests for concurrent DB use, lint, and `just check`. ## Antipatterns - Building SQL with string concatenation and untrusted input. - Global DB handles hidden in packages. - Ignoring `rows.Err()` or leaking `Rows`/transactions. - N+1 queries without measurement or accepted tradeoff. - Tests that depend on a developer's local database. ## Completion Report schema/query changes, transaction model, tool choices, test database approach, commands, and safety boundaries. --- > Source: [nyquistwilder/personal-pi](https://github.com/nyquistwilder/personal-pi) — distributed by [TomeVault](https://tomevault.io).