One thing missing from most local agent setups is identity verification between agents. How do you handle trust when one agent needs to verify another is legitimate before sharing data?