gsd:execute-phase

To execute development phase tasks in parallel waves, you need a phase-plan.md file listing tasks. The system groups independent tasks into concurrent waves, enforces dependency ordering, and generates atomic commits per task.

Wave-based parallel execution is an orchestration method that runs independent development tasks concurrently while respecting dependencies. It speeds up phase completion by grouping tasks into sequential waves separated by verification checkpoints.

Dependency and deviation handling is managed by enforcing strict ordering for dependent tasks. The system applies deviation rules for BUG, MISSING, and BLOCKING incidents, allowing immediate bug fixes or deferring blocking work to backlog.

Yes, you can generate atomic commits per task within a parallel execution plan. The system creates an atomic commit for each completed task to ensure traceability and requests verification checkpoints between execution waves.

The best way to handle blocking incidents in a phase plan is through deviation rules. The system identifies BLOCKING incidents during wave execution and defers the blocked work to the backlog while allowing independent tasks to continue.

You should use wave-based execution when your phase-plan contains independent tasks that can run concurrently. If tasks have strict dependencies requiring sequential processing, the system enforces ordering, but parallel waves maximize efficiency for mixed workloads.