AI governance framework — semantic responsibility, self-auditing memory, council deliberation, and 3,019 tests that prove it. 語魂：讓 AI 對自己說過的話負責。
