What problem does it solve? Large development tasks often fail due to poor decomposition, lost context, inconsistent coding standards, and uncommitted work. This Skill enforces a disciplined N-step loop that captures requirements verbatim, writes detailed specs, delegates to subagents, and guarantees atomic commits with CI/CD monitoring. ## Core Features & Use Cases - Verbatim Prompt Capture & Task Extraction: Records the user's request into .ai-memory/plans/pending/ and outputs a confirmed checklist before any work begins. - Spec-First Orchestration: Splits N steps into a spec-writing phase and an execution phase, with subtask files, per-task agent isolation folders, and strict coding guideline enforcement (boolean naming, semantic tests, AppError handling). - Git Discipline & CI/CD Monitoring: Bans per-file commits, test running, and build checking during turns; mandates a single atomic commit, push, and ETA-based pipeline polling via GitMap. - Use Case: A user asks to implement a new standard UI alongside a modern UI. The Skill records the request, writes a detailed spec with DRY constraints, breaks it into subtasks, executes them across isolated subagents, and ships one consolidated commit. ## Quick Start Ask the AI to execute your parent task broken down into N steps using this orchestration skill, providing the full task description in your prompt.
