帮助 AI 查询团队决策台账，行动前确认历史决策是否仍然有效。
Prevents autonomous agents from fabricating tool results by verifying HMAC-signed receipts against epistemic claim types, forcing re-grounding or escalation before actions commit.
