What problem does it solve? Executing a multi-task implementation plan in a single AI session leads to context pollution, lost progress after compaction, and unreviewed code. This Skill orchestrates plan execution by delegating each task to an isolated subagent with its own review gate, keeping the controller's context clean and every change verified. ## Core Features & Use Cases - Per-task subagent dispatch: Each plan task gets a fresh implementer subagent with a file-based task brief, followed by an independent task reviewer checking spec compliance and code quality. - Bounded fix loop with escalation: Up to five fix rounds per task, resuming the original implementer for rounds 1-3 and escalating to a more capable model for rounds 4-5, with adjudication rulings recorded in a ledger. - Ledger-based recovery: A per-plan workspace with a progress ledger survives context compaction, so completed tasks are never re-dispatched. - Use Case: You have a written implementation plan with eight mostly independent tasks. This Skill creates a git worktree, dispatches implementers and reviewers per task, runs a final whole-branch review, and hands off to branch finishing. ## Quick Start Execute the implementation plan at docs/plans/feature-plan.md using subagent-driven development with per-task reviews.
