stride-subagent-workflow

Subagent orchestration uses a structured decision framework to dispatch decomposition, exploration, planning, and review agents based on task metadata like complexity and acceptance criteria. This ensures tasks are properly sequenced for reliable incremental delivery.

Task decomposition and exploration are needed for medium-to-large tasks with multiple key files or acceptance criteria. The workflow determines when to dispatch each phase to improve code quality and traceability before implementation.

After claiming a task, invoke the subagent workflow to determine which agents to dispatch for decomposition and review. This defines the flow between task claiming, orchestration, and completing tasks with a traceable review record.

Subagent orchestration applies to medium-to-large tasks, goals, or tasks with multiple key files and acceptance criteria. Small single-file tasks likely do not require dispatching the full decomposition, exploration, and review phases.

Enforcing guardrails during multi-agent execution requires coordinating built-in and custom agents through a phase-based workflow. This produces a traceable review record for each task while maintaining proper subagent sequencing.

The subagent workflow coordinates both built-in and custom agents during task execution. It applies the decision matrix to determine when to dispatch decomposition, exploration, planning, and review agents for incremental delivery.