auditing-web-performance

Measure Core Web Vitals and load timing for web pages using Playwright browser sessions.

Updated Apr 11, 2026
One-click install
npx skills add https://github.com/celikgo/webmobai --skill auditing-web-performance-celikgo
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: auditing-web-performance
Source: https://github.com/celikgo/webmobai/tree/main/.claude/skills/auditing-web-performance
Command: npx skills add https://github.com/celikgo/webmobai --skill auditing-web-performance-celikgo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? It answers whether a web page is fast enough by measuring Core Web Vitals (LCP, FCP, CLS, INP, TTI, TTFB) in a real browser session, rating each metric against Google's thresholds, and flagging regressions against baselines or recorded history. ## Core Features & Use Cases - Web Vitals collection: Captures LCP, FCP, CLS (running and at load), TTI (fast and strict), INP, TTFB, and load timings via WebMobAI's performance tools, with per-metric Good / Needs Improvement / Poor ratings. - Multi-run aggregation: Runs a page 1-10 times and reports median, p95, min, and max so quoted numbers survive single-run variance. - Throttled mobile simulation: Applies network presets (slow-3g, fast-3g, slow-4g, offline) and CPU slowdown on Chromium, plus mobile viewports, to approximate real-device conditions. - Use Case: After a deploy, ask whether LCP regressed; the skill measures the page over 5 runs, diffs the median against your baseline, identifies the LCP element via an in-page PerformanceObserver snippet, and generates an HTML report. ## Quick Start Ask the assistant to audit Core Web Vitals for a URL such as "measure performance on https://example.com/pricing and tell me if anything is slow".

Frequently Asked Questions about auditing-web-performance

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I measure Core Web Vitals for a web page?▼

Launch a browser session, navigate to the URL, wait for the page to settle, then collect metrics with the performance tool. It returns LCP, FCP, CLS, TTI, INP, and TTFB rated against Google's Good, Needs Improvement, and Poor thresholds.

How to check if a deploy caused a performance regression?▼

Measure the page with multi-run aggregation (3-10 runs) and compare medians against your baseline, since single-run variance is plus or minus 15-20%. You can also check recorded run history to compare the latest entry against the median of prior runs.

What is the difference between this and a Lighthouse performance score?▼

This produces millisecond-level metric diagnosis from your live browser session, while Lighthouse computes the composite 0-100 score in its own throttled headless Chrome. The numbers will not match; use Lighthouse for the headline score and this for per-metric diagnosis.

Does network throttling work on Firefox and WebKit?▼

No, bandwidth throttling presets are Chromium-only because they drive the Chrome DevTools Protocol and are silently ignored on Firefox and WebKit. The offline preset is the exception, working on every engine via Playwright's context-level switch.

Why is my CLS measurement higher in long sessions?▼

The running CLS value accumulates over the entire page lifetime, so it inflates the longer the page stays open. Use the CLS at load value, which is frozen 3 seconds after the load event, for a faithful arrival-experience number.

Can I measure performance of pages behind a login?▼

Yes, launch the browser with a saved storageState session file to test authenticated pages. Note that Lighthouse cannot do this, so gated pages are limited to raw metric collection rather than the composite score.