benchmark

Web performance regression detection works by collecting navigation and resource timing data via a local browse daemon. It baselines page metrics like TTFB and LCP, then compares new audits against these saved JSON baselines to identify performance drift.

To capture a baseline for web vitals analysis, run a full-page audit on the target URL. This records core metrics like TTFB, FCP, and LCP into a standardized JSON benchmark format, establishing the reference point for future trend comparisons.

Yes, a local browse daemon is required to collect navigation and resource timing data. The daemon gathers the raw timing metrics needed to generate baselines, apply diffs, and output comparison reports.

PR-aware benchmarking compares new timing data against stored baseline JSON files during audits. It calculates diffs to alert on performance drift, enabling automated trend analysis of web vitals across multiple page changes over time.

Yes, you can generate automated reports for web performance health. After collecting timing data and comparing it against baselines, the Skill outputs reports and dashboards to communicate performance metrics to stakeholders.

A limitation of using JSON benchmark formats for regression detection is the strict requirement for pre-existing baseline files and a local browse daemon. Without these prerequisites, the Skill cannot collect timing data or output baseline comparisons.