What problem does it solve? When optimizing the Tesserae UI toolkit or changing a layout container, component, or stylesheet, you need to prove two things: that the change actually reduced build or rendering cost, and that it did not alter what the user sees. Naive pixel or geometry diffs produce misleading results, and this Skill provides the correct measurement and comparison methodology. ## Core Features & Use Cases - CPU profiling and attribution: Capture CDP CPU profiles of the Tesserae.Bench harness app, analyze self and inclusive time by function and file, and walk the call tree to attribute hot frames to the responsible component. - A/B build comparison: Interleave benchmark runs of two builds, take best-of measurements, and honestly report build-cost wins versus session wall-clock noise. - Rendering verification gates: Choose the right comparison method per change type — geometry comparison, text-run diffing across 131 gallery samples, or pixel diffing with visual diff images — and establish a noise floor via self-comparison. - Use Case: After removing a wrapper element from Stack layout, run textdiff-samples.js against both builds to confirm text runs did not move horizontally, then use inspect.js to attribute any remaining differences to a specific CSS property. ## Quick Start Ask the AI to benchmark the current Tesserae build against the previous one and verify that the layout change did not alter rendering.
