benchmark

You can detect web performance regressions by capturing per-URL performance baselines and comparing current page metrics against them. This Skill measures TTFB, FCP, LCP, and full load timings to identify regressions across single or multiple pages.

Yes, you can view performance benchmark trends by generating diff and trend reports. The Skill saves these reports as JSON or Markdown files under the .gstack/benchmark-reports directory, allowing you to track page metrics across branches over time.

Establishing a baseline requires access to the browser performance API and the browse daemon's perf tool. It captures key web performance metrics including TTFB, FCP, LCP, DOM Interactive, DOM Complete, and resource analysis timings.

Run a quick performance check against a stored baseline by executing the benchmark command with your target URL and the --diff flag. This compares current real browser timings directly against your previously captured per-URL baseline.

Yes, this Skill provides automated and reproducible workflows for benchmarking page performance across PRs. By capturing baselines and comparing real browser timings, it ensures consistent regression detection as application changes accumulate.

Capturing performance baselines requires the browse daemon because it provides the perf tool and access to the browser performance API needed to gather real browser timings. This ensures accurate measurement of TTFB, FCP, and LCP metrics.