What problem does it solve? Turning an approved discovery into an implementable specification is often where requirements become vague, unverifiable, or untraceable. This Skill converts a discovery into a structured spec that planners can implement, QA reviewers can verify, and release steps can expose, with every requirement traced back to a discovery need and an acceptance criterion. ## Core Features & Use Cases - Traceable Requirements: Generates numbered functional and non-functional requirements, each verifiable and tied to a discovery need, with thresholds instead of adjectives (e.g., "< 200ms p95" instead of "fast"). - Acceptance Criteria & Test Plan: Produces given/when/then acceptance criteria per requirement plus a test plan covering unit, integration, and e2e scope, including what is excluded and why. - Phased Scoping: Enforces environment constraints (3-5 files per change) and splits oversized work into mergeable phases: minimal, happy path, edge cases, optimization. - Use Case: After approving a discovery for a new checkout feature, invoke /spec to produce the full specification covering data model changes, dependencies, security surfaces, and acceptance criteria, then approve it before implementation begins. ## Quick Start Ask the AI to run /spec on the approved discovery document to generate a traceable specification with requirements, acceptance criteria, and a test plan.
