What problem does it solve? Starting to code from vague or ambiguous requirements leads to rework, wrong assumptions, and features nobody asked for. This Skill forces a written specification to be created and reviewed before any implementation begins, making the spec the shared source of truth between the AI agent and the human engineer. ## Core Features & Use Cases - Gated Four-Phase Workflow: Enforces a Specify → Plan → Tasks → Implement pipeline where each phase requires human review before advancing. - Structured Spec Template: Covers six core areas including objective, commands, project structure, code style, testing strategy, and Always/Ask First/Never boundaries. - Assumption Surfacing: Requires listing assumptions and reframing vague requirements into concrete, testable success criteria before writing code. - Use Case: When asked to "build a dashboard feature" with unclear requirements, the agent first drafts a spec with acceptance criteria (e.g., LCP < 2.5s), gets human approval, then plans and implements against it. ## Quick Start Ask the agent to write a specification for the new feature before writing any code, covering objectives, commands, structure, testing, and boundaries.
