What problem does it solve? Turning a raw engineering request into an approved, implementation-ready design is error-prone: requirements stay vague, current-state behavior goes uninvestigated, and design decisions lack evidence. This Skill structures that entire upstream workflow so designs are grounded in verified evidence and pass architecture review. ## Core Features & Use Cases - Mandatory Core Artifacts: Generates a requirements doc, investigation notes, and a design spec from structured templates, plus a solution revision record tracking every review round. - Evidence-Backed Investigation: Records exact sources, commands, runtime probes, and stable behavior IDs so downstream reviewers never rediscover facts. - Design Health & Migration Decisions: Classifies root causes, decides refactor posture, and makes explicit persisted-data transition decisions (no migration, rebuild, or migration plan). - Use Case: A team receives a feature request touching an existing codebase. The Skill bootstraps a dedicated git worktree, investigates current behavior, refines requirements with the user until approved, produces a spine-and-ownership design spec, and iterates with the architecture reviewer until the design passes. ## Quick Start Ask the agent to investigate this task and produce the requirements doc, investigation notes, and design spec for architecture review.
