What problem does it solve? Performance regressions accumulate silently in a codebase because nobody schedules time to hunt for them. This Skill runs a structured, repeatable performance review that surfaces the ten highest-value improvements, scores them by impact and effort, and turns the best one into a reviewed, tested pull request. ## Core Features & Use Cases - Scored Findings Scan: Analyzes the codebase against a performance checklist covering allocations, async patterns, LINQ and collections, and I/O or database access, then ranks the top 10 findings by impact and effort. - TDD-Based Implementation: Implements the winning finding through a Red-Green-Refactor cycle with a failing benchmark or test, followed by a code review before committing. - Gated Pull Request: Presents the findings table and fix for explicit human approval before opening a labeled PR, keeping automated changes reviewable. - Use Case: Run weekly on a .NET service to steadily eliminate issues like repeated LINQ enumeration or sync-over-async database calls, one focused PR at a time. ## Quick Start Run the performance review on the default scope and open a PR for the highest-impact, lowest-effort finding.