orchestrator-agent

Expert routing in Nebuchadnezzar v3.5 uses keyword- and stage-based routing to coordinate the right expert for each task. The orchestrator-agent applies this routing across Expert CRUD, Observability, ADHD Execution Loop, and Learning Propagation, moving work via structured YAML handoffs between experts.

Red stage signals trigger observability-driven routing decisions within the orchestrator-agent. The system identifies the red condition, loads the recommended expert, and passes a structured handoff payload to resolve the issue and restore normal pipeline operation.

Learning propagation occurs after sessions complete via EOD synchronization, triggered by the orchestrator-agent. Expertise improvements flow end-to-end through the pipeline, ensuring each expert benefits from prior session outcomes and refines future decisions.

Session startup protocol is enforced by the orchestrator-agent as part of the Four Pillars governance. It initializes new sessions with proper expert assignment, routing rules, and handoff structures to ensure consistent coordination from the beginning of each workflow.

Yes, handoff-oriented workflows are core to the orchestrator-agent. Work moves between experts using YAML payloads that structure the handoff data, enabling multi-expert coordination across routing, observability, execution, and learning tasks.

The Four Pillars governance is the framework the orchestrator-agent enforces to align expert coordination. It provides structured rules for routing, session management, observability signals, and learning propagation across the entire Nebuchadnezzar v3.5 Pipeline OS.