executing-plans

The parallel-session alternative to sp-subagent-driven-development: load the plan, review it critically for concerns, create todos, then execute every task in order — following each bite-sized step exactly and running its verifications — with human review at natural checkpoints. When all tasks pass, it hands off to sp-finishing-a-development-branch.

The skill is explicit that it is the fallback: “Superpowers works much better with access to subagents… if subagents are available, use sp-subagent-driven-development instead.” It is for harnesses without subagent support, or for running the plan in a separate session from the one that wrote it. Guardrails: stop and ask on any blocker rather than guessing; never start implementation on main/master without explicit consent; return to plan review if the approach needs rethinking. Executors follow sp-test-driven-development per task → artifact-atomic-commit.

It is the seventh-plus member of the cross-framework execute clusteropenspec-applyaddy-incremental-implementationgsd-execute-phasespeckit-implementce-workbmad-dev-storymp-implement — the “walk the plan task-by-task” loop; its distinctive flavor is batch execution with human checkpoints (vs subagent-driven’s continuous autonomous run).

See Also