execute-plan

Plan execution automates moving through each phase of a development plan stored in `.llms/`, delegating work to specialized subagents and tracking progress automatically. The Skill identifies your active plan, runs phase-specific tasks, updates status at each boundary, and commits changes to maintain an auditable trail.

Yes, plan execution detects in-progress plans in `.llms/` and resumes from the current phase. If multiple active plans exist, you choose which to continue, then the Skill proceeds with phase-specific subagent delegation and progress tracking from that point forward.

Plan execution detects validation failures and orchestrates fix cycles automatically. It creates structured subplans to address issues, reruns validation, and continues phase progression only after prerequisites are met, maintaining strict phase boundaries throughout.

Plan execution commits changes at each phase boundary, creating a clear auditable history linked to phase completion. This ensures every phase transition is recorded and traceable, supporting both progress visibility and rollback capability if needed.

No, plan execution automatically spawns phase-specific subagents—scope analysis, context gathering, solution design, and others—based on your plan structure. Each subagent handles its phase's work and reports findings back to update the plan.

Plans must be stored in `.llms/` directory and structured with distinct phases. Plan execution expects phase definitions, prerequisite markers, and status fields so it can track progress, delegate to subagents per phase, and manage validation cycles consistently.