Agent audit logs serve fundamentally different purposes than traditional application logs. While application logs answer operational questions about system failures, agent audit logs must capture authorization chains, agent identity, delegation records, and tool invocations to establish accountability. Existing logging infrastructure is insufficient because it was not designed for autonomous systems that act on behalf of users across extended sessions.