benchmark

Web performance regressions are detected by collecting navigation timing, resource timing, and bundle size data, then comparing current metrics against historical baselines. This Skill automates that measurement and comparison across PRs to surface regressions and trends.

Performance baseline analysis captures TTFB, FCP, LCP, DOM Interactive, DOM Complete, and Full Load metrics. It establishes baselines by collecting these navigation and resource timing measurements, enabling automated comparison to detect degradation over time.

To run a quick performance check on a specific page route, execute the benchmark command with the target URL. This initiates a full performance audit, collecting timing data and comparing results against the stored baseline to surface any regressions or warnings.

Yes, a browse daemon is required to collect web vitals for trend analysis. The daemon gathers the navigation timing and resource data, which is then used to generate JSON and MD reports highlighting historical degradation or improvement patterns across baselines.

Performance regression and trend reports are output in both JSON and Markdown formats. These reports summarize regressions, warnings, slowest resources, and historical trends, writing the captured baselines to the designated benchmark-reports directory.