The agent that produced this output chunk. Use this to apply agent-specific policies.
The agent running this guardrail (the observer/supervisor).
Optional agencyAgency ID if both agents belong to the same agency. Useful for agency-level policy enforcement.
The output chunk from the source agent.
Guardrail context for policy decisions.
Payload for cross-agent output evaluation.
Provides information about both the source agent (producing output) and the observer agent (running the guardrail).