What problem does it solve? Codebases accumulate shallow modules — thin wrappers, pass-through functions, and single-use helpers whose interfaces are nearly as complex as their implementations. This makes code hard to test, hard for AI agents to navigate, and spreads bugs across many files. This Skill audits a codebase for these friction points and proposes concrete deepening refactors that consolidate logic behind small, testable interfaces. ## Core Features & Use Cases - Architectural audit: Explores the codebase (weighted toward recently changed hot spots), applies the deletion test, and surfaces 3–8 deepening candidates classified by dependency category (in-process, local-substitutable, ports-and-adapters, mock). - HTML report generation: Renders candidates as a self-contained HTML report with Mermaid before/after diagrams via a Jinja2-based Python script, opened automatically in the browser. - Multi-language idiom packs: Ships idiom guides for Python, Swift, TypeScript, and React that map the architecture vocabulary (module, interface, seam, adapter) to native constructs like Protocol, protocol, interface, and hook seams. - Guided design loop: After the user picks a candidate, runs a grilling conversation, updates CONTEXT.md domain glossaries, offers ADRs for rejected candidates, and can spawn parallel sub-agents to design alternative interfaces. - Use Case: Before letting AI agents loose on a legacy service, run the audit to find tightly-coupled modules leaking across seams, review the HTML report, then deepen the top candidate into one interface with one test surface. ## Quick Start Ask the assistant to review this repository for shallow modules and produce an architecture report of deepening opportunities.
