_evaluation

Multi-perspective evaluation coordination selects diverse perspectives and spawns separate evaluators with task-specific criteria, aggregating independent verdicts to avoid bias and routing findings for user decision making.

An evaluation orchestrator applies stage-aware criteria by assigning task-specific rules to spawned evaluators at each workflow stage, then filtering aggregated results through confidence and severity scoring thresholds.

Yes, evaluation workflows support scoring with confidence and severity, enforce threshold filtering on aggregated verdicts, and route only findings meeting defined criteria to the user.

Aggregating independent verdicts requires an orchestrator to coordinate parallel evaluators, collect their separate judgments, apply stage-appropriate criteria, and consolidate findings for routing to the user.

Evaluation governance is needed when workflows require multi-perspective judgment and user control to avoid bias, ensuring independent verdicts are routed to the user for final decision making.

Yes, the evaluation orchestrator provides integration guidance for reporting after aggregating independent verdicts and enforcing threshold filtering on confidence and severity scores.