Parameters

The package is parameterized so it depends on no app’s data model. Parameters are supplied once, at import time, via imports[].params:

Parameter	Default	What it controls
user_table	users	FK target for created_by / updated_by and audit_log.changed_by
user_pk	user_id	the primary-key column of user_table the FKs point at
actor_guc	app.actor_id	the GUC the audit triggers read the actor id from
lenient_guc	app.audit_lenient	GUC that, when 'true', lets audit_stamp tolerate a missing actor (bootstrap seeding) instead of raising

All four default to the convention, so the common case is param-free:

imports:
  - package: '@smplcty/schema-std'

Overriding

Override any subset; the rest keep their defaults. One override repoints created_by, updated_by, and audit_log.changed_by at the same identity table, because the audit and audit_log_actor mixins share user_table / user_pk:

imports:
  - package: '@smplcty/schema-std'
    params:
      user_table: accounts
      user_pk: account_id
      actor_guc: app.account_id

Where each parameter lands

• user_table / user_pk are interpolated into the FK references of the audit mixin’s created_by / updated_by columns and the audit_log_actor mixin’s changed_by column:

  references:
    table: '{{user_table}}'
    column: '{{user_pk}}'
    on_delete: RESTRICT
    on_update: CASCADE

• actor_guc is interpolated into the function bodies of audit_stamp, audit_diff, and audit_backfill_by, where the actor id is read:

  NULLIF(current_setting('{{actor_guc}}', true), '')::bigint

• lenient_guc is interpolated into audit_stamp, which raises unless this GUC is 'true' when the actor is missing:

  current_setting('{{lenient_guc}}', true) IS DISTINCT FROM 'true'

The actor type

created_by / updated_by / changed_by are bigint, and the GUC value is cast to bigint. The user_pk column you point at must be an integer key.