What problem does it solve? It turns an approved technical design into working source code without silent design drift, enforcing layer order, traceability tags, and quality gates so every implemented scenario stays reviewable and auditable. ## Core Features & Use Cases - Design-driven implementation: Reads the technical design document and component map, then creates or modifies exactly the files the design specifies, stopping when the design is under-specified or conflicts with the codebase. - Layered, traceable code generation: Builds in the stack profile's layer order and tags entry points with @trace.implements annotations linking code back to use case scenarios. - AI prompt registry handling: Creates new versioned prompt YAML files for AI features without editing published versions, wiring call sites through a mockable LLM protocol. - Use Case: A tech lead approves a design for a three-use-case feature; the skill fans out one code-implementer agent per use case, runs lint/typecheck/build checks, and records the G4 gate verdict. ## Quick Start Invoke /gen-code with a feature or use case ID, for example: implement FEAT-042 from its technical design document.
