execute-phase

To coordinate multi-phase project execution, you can use wave-based orchestration to load project state, discover plans, and spawn subagents to execute tasks in parallel waves while tracking dynamic state. Checkpoints ensure recoverability between these execution waves.

Wave-based plan execution is an orchestration method where an orchestrator delegates plan discovery and dependency analysis to subagents, distributing tasks across parallel execution waves rather than running sequentially. It aggregates per-plan summaries into a phase-wide report.

To save progress and manage dynamic state between execution waves, the orchestrator tracks progress and saves decisions at checkpoints. This ensures recoverability between waves and maintains phase-state integration with planning artifacts and version control.

Yes, you can execute multiple plans in parallel across subagents even when dependencies exist. The orchestrator performs dependency analysis to determine which plans can run concurrently within a wave, ensuring subsequent waves respect prior results and checkpoint states.

You should use wave-based orchestration instead of sequential execution when your multi-phase project contains plans that depend on prior results but can be grouped into parallel waves. It maximizes throughput by spawning subagents for concurrent execution where dependencies allow.

No, you do not need to manually load project state before running phase execution. The orchestrator automatically loads project state, discovers available plans, and coordinates the distribution of tasks to subagents to ensure progress across waves.