pairing

What problem does it solve? When an AI builds an entire feature, document, or plan in one uninterrupted run, the person only gets to review a finished result, which makes course corrections expensive and lets errors pass unreviewed. This Skill paces collaborative work into reviewable pieces so the person stays in the lead and steers while the work happens. ## Core Features & Use Cases - Concern Splitting and Sorting: Splits a request into separate concerns (never bundling code changes with questions), then sorts each as skill-backed, decision, prose, or open-ended work. - Plan-Before-Build: Proposes a plan naming concerns, pieces, expensive-to-reverse choices, and the feedback record location, then waits for approval before building anything. - Stop-and-Review Loop: Builds one piece at a time, presents what can be checked, ends the turn, and routes feedback to fix, carry forward, or revise the plan. - Backing Skill Orchestration: Delegates test-first builds to tdd, restructuring to refactor, interface contracts to design-an-api, and plan work to iterative-plan-review or plan-implementation, each run collaboratively. - Use Case: Ask to pair on building a new API endpoint; the Skill proposes a plan, hands the test-first build to the tdd skill one unit at a time, and stops after each piece so you review the work as it happens. ## Quick Start Ask the AI to pair with you on building a feature, making a design decision, or drafting a document, and it will propose a plan and build it piece by piece with your review at each stop.