wave-execution

Parallel task execution minimizes cycle time by grouping tasks into dependent waves and dispatching independent tasks to parallel workers. The engine analyzes your plan, builds a dependency graph, and derives wave boundaries to maximize throughput while preserving correctness.

Dependency mapping for task scheduling works by analyzing a multi-step plan to build a dependency graph and derive wave boundaries. It assigns tasks to sequential or parallel waves to optimize concurrency and ensure dependent integration tasks wait for setup tasks to finish.

Coordinating parallel workers without conflicts is best achieved by computing wave boundaries from a dependency graph and executing isolated task contexts in sequential waves. Post-wave verification checks ensure tasks complete and validate integration between parallel tasks.

Wave scheduling supports multi-component feature plans by running independent setup tasks in parallel while dependent integration tasks wait. It applies to any multi-step plan where tasks have clear dependencies and can be dispatched to subagents or parallel workers.

You should avoid parallel wave execution when a plan lacks clear dependencies between tasks or when tasks cannot be safely dispatched to isolated subagent contexts. It is specifically designed for multi-step plans requiring dependency management and post-wave integration checks.