What problem does it solve? Teams shipping large volumes of AI-generated code accumulate "dark code" — code no human ever understood — creating hidden systemic risk that surfaces during incidents, security reviews, or regulatory scrutiny. This Skill diagnoses where that risk concentrates before remediation begins. ## Core Features & Use Cases - Structured Context Gathering: Interviews the user in four staged groups covering system architecture, AI tool usage, team ownership, and deployment practices. - Two-Dimensional Risk Analysis: Classifies hotspots as structural dark code (emergent agent-assembled behavior) or velocity dark code (authored code nobody comprehended), plus compounding factors like ownership gaps and talent erosion. - Prioritized Risk Report: Produces a hotspot severity map, top-3 failure scenarios, ownership gap list, comprehension debt scorecard, and an ordered action plan. - Use Case: Before a security review or board-level risk assessment, run the audit quarterly to identify which services have critical comprehension gaps and which remediation actions deliver the highest impact. ## Quick Start Use the dark-code-audit skill to assess where dark code is accumulating in our system and prioritize what to fix first.
