What problem does it solve? Turning a written implementation plan into working code requires coordinating many steps—reading task docs, isolating work, executing phases in order, tracking progress, and opening a PR—which is error-prone when done manually or in a single unstructured agent session. ## Core Features & Use Cases - Coordinator-Subagent Execution: Dispatches one subagent per plan phase, following the Execution Sequence DAG from plan.md, with serial or parallel execution as the DAG allows. - Isolated Worktree Workflow: Creates a git worktree with branch task/$ID-$NAME so implementation never touches the main planning branch. - Progress Tracking & Resumption: Updates plan.md checkboxes and commits after each phase, so an interrupted session resumes from the first unchecked phase. - Use Case: After committing planning docs for task 042, run the skill with the task ID; it reads the plan, executes each phase via subagents, marks the task complete, and opens a PR with a structured body. ## Quick Start Execute task 042 by reading its planning docs and implementing each phase in an isolated worktree, then open a pull request.
