What problem does it solve? In repositories where agents generate code at high throughput, architectural constraints that are not mechanically enforced get violated quickly — circular dependencies, cross-layer imports, and unparsed external data creep in. This Skill establishes layered architecture rules and audits them with a read-only boundary-auditor agent. ## Core Features & Use Cases - Layering Model Design: Derives a project-specific dependency direction (e.g., Types → Config → Repo → Service → Runtime → UI) with cross-cutting concerns funneled through a single Providers chokepoint, documented in docs/ARCHITECTURE.md. - Mechanical Boundary Audits: The boundary-auditor agent uses Grep/Bash to detect circular dependencies, cross-layer violations, and data boundary violations, producing a severity-classified report (CRITICAL/HIGH/MEDIUM/LOW) with actionable fix instructions. - Parse, Don't Validate Enforcement: Requires external data entering system boundaries to be parsed into strong types rather than passed around as any/dict. - Use Case: A team notices Service-layer files directly importing Repository implementations. Run this Skill to confirm the dependency rules, grep for violations, and receive a report with file line numbers and concrete refactoring steps. ## Quick Start Ask the agent to audit this repository for cross-layer dependency violations and circular dependencies, then generate an architecture boundary report with fix suggestions.
