SOUL.md compliance: 'Every response I generate should be logged locally
with the inputs that produced it, the sources I consulted, and the
confidence assessment I made.'
scripts/audit_trail.py:
AuditTrail class with log_response(), query(), get_stats(), get_by_session()
Content-addressed IDs, JSONL storage, local-only (never sent anywhere)
User can query: 'why did you say X?' -> search trail for matching entries
tests/test_audit_trail.py: 7 tests passing
