orchestration

You can coordinate parallel AI workflows by decomposing complex tasks into concurrent subtasks, tracking dependencies with cc-mirror, and monitoring live session progress via TodoWrite. This approach turns ambitious goals into actionable, observable outcomes with minimal friction.

Decomposing complex tasks into parallel subtasks is best handled by an orchestrator that establishes persistent dependencies and spawns worker agents. It uses a two-layer task management model to scale orchestration across workers and synthesize their results.

cc-mirror tracks dependencies in multi-agent orchestration by maintaining persistent task states across concurrent worker agents. This two-layer task management model ensures that complex workflows scale correctly while preserving the execution order of dependent subtasks.

Yes, you can use TodoWrite for live session progress updates in parallel task management. It provides real-time visibility into worker status and task completion, allowing you to observe multi-agent workflows as they synthesize actionable outcomes.

You need domain references to structure multi-agent workflow execution because they provide the necessary guides and patterns for domain-guided execution. Reading these references ensures that spawned workers correctly synthesize results according to established structural patterns.

You should not use a multi-agent orchestrator for task management if your work cannot be decomposed into concurrent subtasks with persistent dependencies. It is designed for scaling parallel workflows and synthesizing results, making it unsuitable for strictly sequential or non-dependent tasks.