parallel-task

You orchestrate parallel subagents by parsing a development plan, resolving task dependencies, and launching waves of subagents to execute tasks in the correct order with traceable progress logging.

Wave-based task execution is a coordination method that launches parallel subagents in sequential groups, ensuring tasks within each wave run concurrently while respecting dependency-driven ordering across waves.

You resolve task dependencies by parsing the plan document to identify prerequisite relationships, then grouping tasks into waves so that dependent tasks only execute after their dependencies complete.

Yes, you can provide a plan file path along with an optional task subset, and the orchestrator will execute only those specified tasks while still respecting dependency-driven wave ordering.

The best way to ensure traceability is to update plan documents with execution logs after each completed task, providing atomic commits and explicit logging for every subagent action.

Orchestrated task execution fails when dependency resolution is incomplete, so the orchestrator must strictly parse plan dependencies to guarantee that no wave launches tasks before their prerequisites are fully committed.