ORM Adapters

Go Audit supports ORMs that expose lifecycle hooks — that's the primary requirement for automatic tracking. Three adapters ship with the project, each in its own sub-module.

Comparison Table

All three deliver the same stored record shape; the integration plumbing is what differs.

Why No sqlx / pgx / database/sql?

Thin SQL wrappers like sqlx, pgx, and database/sql don't have a lifecycle model for create/update/delete — they just execute SQL. Go Audit needs a hook point to capture old values before an update and compare them after, which these libraries don't provide.

If you need auditing for a raw-SQL code path, use the manual entry point directly:

_ = auditor.RecordDataChange(ctx, audit.DataEntry{
    EntityType: "ledger_entries",
    EntityID:   "le-42",
    Action:     audit.ActionCreate,
    NewValues:  map[string]any{"amount": 500},
})

See the Data Audit API for the full DataEntry shape.

Choosing an Adapter

• GORM — most popular, easiest setup, largest plugin ecosystem. Has the most complete snapshot coverage (including soft-delete auto-detection). Default choice for new projects.
• Bun — lightweight, explicit query builder feel. Great for SQL-ish code with strong typing. Caveat: snapshots run on the base *bun.DB, not within *bun.Tx, so uncommitted writes in the same transaction aren't visible.
• Ent — schema-first code generation with rich graph traversal. Best if you already build models via ent codegen. Bulk update/delete operations get one audit row per affected ID, but without old_values (ent doesn't expose pre-mutation state for bulk ops).

Links

• GORM adapter
• Bun adapter
• Ent adapter