What problem does it solve? Developers often inherit code written by an agent, a predecessor, or an upstream author without ever seeing the design decisions behind it. This Skill transfers that design into your head at a depth where you can justify it: what the public surface is, how one real call flows through it, and which parts you could not defend if asked. ## Core Features & Use Cases - Public Surface First: Reads exported symbols, CLI entry points, README usage, and tests before touching implementation, grounding every claim in file:line references. - End-to-End Trace: Walks the single most-used path from entry to result, plus git churn analysis to show where the design is still moving. - Diagnostic Teaching: Asks one targeted question at a time to locate the gap between what you believe and what the code actually does, then routes defects (unjustified abstractions, unenforced invariants, doc/code mismatches) to the right follow-up skill. - Use Case: You inherit a library partly written by an AI agent and need to evaluate it as a dependency. The Skill walks you through its public API, traces the main call path, flags vendored or ported code, and hands you a list of defects the code failed to explain about itself. ## Quick Start Ask the agent to walk you through this library, explaining its public API, how the main call path works, and which abstractions you could not justify if asked.
