What problem does it solve? Large feature requests and specs are hard to start, easy to scope-creep, and often get implemented without tests or verification. This Skill turns a specification into granular, verifiable units of work with explicit test-first steps and a mandatory human approval gate before any code is written. ## Core Features & Use Cases - Atomic Task Decomposition: Breaks objectives into tasks touching roughly five files or fewer, each with a RED test spec, GREEN implementation target, and an exact verify command. - Execution Strategy Selection: Declares either Standard Single Agent or Sequential Subagents execution, with per-slice adversarial review gates in heavy mode. - Human Approval Gate: Produces a reviewable plan document (e.g. plan.md in the make-feature worktree) and pauses until the engineer explicitly approves it. - Use Case: Given a spec for a new API endpoint, generate an ordered checklist of TDD tasks where each task names the failing test to write, the minimal implementation, and the exact pytest command to verify it. ## Quick Start Ask the agent to run /plan on your feature specification to produce an ordered, test-driven task breakdown for approval before coding begins.
