superpowers-executing-plans

Executing a multi-step development plan involves loading the plan file, critically reviewing it for blockers, following each step exactly, performing verifications, and marking tasks completed before finishing the branch.

Coordinating plan execution with subagents and git worktrees involves loading a written plan, executing tasks step-by-step with status tracking, and integrating with sub-skills to complete the development branch.

Yes, you can critically review a development plan before starting task execution to surface any concerns or blockers with your human partner and ensure the plan is safe to proceed.

You can track task status during plan execution by marking each task as in_progress before starting, performing the required verifications, and then marking it completed to maintain an auditable execution flow.

After all development plan tasks are verified and completed, the workflow announces the transition to finishing the development branch and initiates the finishing-a-development-branch sub-skill to complete the work.

You should not use automated plan execution when a written plan file is unavailable or when tasks lack clear verification steps, as the workflow requires critical review and explicit status tracking to function correctly.