This document defines the HJS: An Accountability Layer for AI Agents v0.4, an event recording layer designed for AI decision-making systems. HJS implements strict traceability and immutability of AI machine behavior, with configurable human identity protection, enabling AI decision chains that are cryptographically verifiable and fully auditable. HJS does not assign legal liability, enforce monitoring, or manage authorization distribution. It provides only verifiable records of AI behavior and optional privacy protection for human participants to balance transparency with individual rights.