subagents-orchestration-guide

Orchestrate subagents by using a requirement-analyzer to assess task scope, then route work to specialized subagents based on complexity scale. The orchestrator coordinates all execution phases, enforces approval checkpoints, and prevents direct subagent-to-subagent communication, ensuring structured delegation across documentation, design, and testing deliverables.

Use orchestration for multi-phase implementations requiring autonomous execution with human approval gates—such as large-scale feature rollouts, authentication system redesigns, or projects generating PRDs, ADRs, and test plans. Scale-based delegation automatically determines document requirements and stop points, eliminating manual coordination overhead.

Yes. Subagent orchestration automates requirement analysis, scale determination, and task delegation to specialized agents like technical designers and task executors. The system enforces approval workflows between phases, ensuring human oversight while removing coordination friction in multi-agent implementation flows.

Generate PRDs, architecture decision records (ADRs), design documentation, work plans, and test skeletons through coordinated subagent execution. The orchestrator routes each deliverable phase to appropriate agents based on task complexity, with built-in approval checkpoints before progressing to downstream work.

Scale-driven delegation analyzes task complexity and automatically selects appropriate subagents and document requirements. The orchestrator maintains a centralized routing layer that prevents agents from communicating directly, ensuring all decisions flow through the orchestrator for consistent, auditable task progression.

The orchestration framework enforces stop points for user approval between phases, maintains a centralized coordinator that prevents subagent-to-subagent calls, and uses requirement analysis to set proper scale expectations. These safeguards ensure autonomous execution remains auditable and user-controlled.