What problem does it solve? Code reviews often miss the simplest improvement: deleting unnecessary code. This Skill reviews a diff specifically for over-engineering, identifying reinvented standard library functions, needless dependencies, and speculative abstractions so the diff can get shorter instead of longer. ## Core Features & Use Cases - Over-Engineering Detection: Finds dead code, hand-rolled utilities the standard library already ships, and dependencies that duplicate native platform features. - Tagged One-Line Findings: Each finding uses a strict format with tags like delete:, stdlib:, native:, yagni:, and shrink:, naming the exact replacement. - Net-Line Scoring: Ends every review with a net: -<N> lines possible. metric, or reports Lean already. Ship. when nothing can be cut. - Use Case: Paste a pull request diff that adds a 27-line email validator class, and receive a finding like L12-38: stdlib: 27-line validator class. "@" in email, 1 line, real validation is the confirmation mail. ## Quick Start Ask the AI to review the current diff with ponytail-review and list everything that can be deleted or replaced with standard library equivalents.