orchestrator

Multi-RFC orchestration decomposes work into sequential phases, assigning independent sub-agents to each feature area. This prevents scope drift by using PRD and RFC boundaries to verify integration after every phase completes.

Sub-agent orchestration uses a central coordinator to spawn focused agents that write code for independent features. It enforces phase planning and end-to-end verification across 3+ parallel tasks using PRD and RFC scope boundaries.

Yes, a PRD and RFCs are required before orchestrating parallel tasks. They define the scope boundaries that prevent sub-agents from drifting, ensuring focused execution and correct end-to-end integration across the codebase.

Prevent scope drift by using PRD and RFC documents to establish strict scope boundaries. The orchestrator enforces these boundaries by providing context briefs and verifying integration after each phase completes.

Use the orchestrator pattern for large implementations spanning 3+ independent features or RFCs. It acts as a planning and verification layer, spawning sub-agents to write code so you can focus on integration rather than implementation.

Verify end-to-end integration by enforcing phase planning and generating verification checklists after sub-agents complete their tasks. The orchestrator produces integration reviews to ensure correctness across all parallel feature areas.