What problem does it solve? Raw benchmark output from just benchmark is hard to act on: numbers without context don't tell you whether a result is trustworthy, whether scaling is acceptable, or whether a regression actually occurred. This Skill provides a structured decision process for reading XCTest performance measurements and deciding what to do next. ## Core Features & Use Cases - Reliability Assessment: Classifies results by relative standard deviation so you know whether a measurement can support a conclusion before comparing numbers. - Scaling Analysis: Computes 1x vs 2x ratios to distinguish constant, linear, super-linear, and quadratic behavior in GRDB database operations. - Anomaly Detection: Interprets per-iteration value arrays to spot outliers, steady growth from database leakage, and bimodal distributions. - Use Case: After a code change, a benchmark fails against its Xcode baseline. Use this Skill to confirm the regression, check stddev validity, then use Instruments with os_signpost to locate the bottleneck before optimizing. ## Quick Start Run just benchmark, save the output, and ask the assistant to interpret the results and recommend next steps.
