orchestrate

Multi-agent orchestration coordinates specialist agents by splitting broad tasks into parallel or sequential sub-tasks. It passes explicit context between agents and synthesizes final results in the main session to complete interdependent workflows efficiently.

Use multi-agent orchestration when a task is too broad, interdependent, or time-sensitive for a single agent. It is ideal for staged analysis, role-based task division across domains like frontend and backend, and parallel research workflows.

You split work by dependency or expertise, assigning specialized areas to different agents. Sequential workflows pass exploration outputs into planning and review, while parallel workflows run independent investigations simultaneously and merge the outcomes.

Yes, you can use agent orchestration to investigate two architectural options at the same time. By spawning parallel sub-agents for each option, you can synthesize the findings into a recommendation with clear tradeoffs.

Subagent spawning requires explicit context passing and main-session orchestration to synthesize final results. Complex interdependencies may create overhead if tasks cannot be cleanly split by dependency or expertise.

No specific dependencies are required to manage staged analysis and review. The orchestrate skill operates independently to structure task decomposition, pass context between phases, and synthesize multi-domain results.