Name of the agent that was invoked.
Input prompt or message sent to the agent.
Final text output produced by the agent.
Ordered list of tool invocations made during this call.
Tool name.
Arguments supplied by the model.
Optional result?: unknownReturn value from the tool (present on success).
Optional error?: stringError message if the tool failed.
Optional guardrailGuardrail evaluation results for this agent call.
Guardrail identifier.
Whether the guardrail check passed.
Action taken by the guardrail (e.g. "allow", "block", "redact").
Token usage for this individual agent call.
Optional costUSD?: numberCost in USD for this call, when available.
Wall-clock milliseconds for this agent call.
Optional emergentWhether this agent was synthesised at runtime by the emergent subsystem.
A complete record of a single agent invocation within an agency run. Appended to
GenerateTextResult.agentCallsand surfaced inApprovalRequest.context.