Content
78%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A highly actionable, well-structured skill body with excellent progressive disclosure and copy-paste-ready configuration and CI examples. Its main weakness is token efficiency: several sections explain Web Vitals fundamentals and lab-vs-field concepts Claude already knows, which could be trimmed to a lean reference of the skill-specific config, thresholds, and anti-patterns.
Suggestions
Compress the Overview to the metric-threshold table plus one line on the 75th-percentile standard; drop the INP-replaced-FID history and 'what each metric measures' prose, since Claude already knows Core Web Vitals.
Cut the 'Lab vs field' section to a two-row table (or merge it into When to use) and the Mobile-vs-desktop prose to just the dual-config pattern and run commands.
Add an explicit validation checkpoint before CI wiring — e.g., 'run `npx lhci autorun` locally once and confirm all error-level assertions pass before committing the workflow' — plus one line on reading a failed assert report.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body re-explains concepts Claude already knows across several sections — the Overview's Core Web Vitals primer ("Google's three canonical user-experience metrics", INP-replaced-FID in 2024, 75th-percentile measurement), "mobile is consistently slower... on a less-powerful CPU over a less-stable network", and the Lab-vs-field distinction. Mostly efficient otherwise, but the padding spans multiple sections rather than isolated instances. | 3 / 5 |
Actionability | Fully executable throughout: install command, a complete copy-paste-ready `.lighthouserc.js` with commented fields, `lhci autorun` plus per-phase and per-URL invocations, assertion-level semantics, and a complete GitHub Actions workflow. The only placeholder (`https://github.com/...`) is an environment-specific value. | 5 / 5 |
Workflow Clarity | Install → Configure → Running → CI integration is clearly sequenced, the three autorun phases are enumerated, and the anti-patterns table provides a promote-warn-to-error feedback loop. Missing an explicit validation checkpoint (e.g., run `lhci assert` locally once before wiring CI) and failure-recovery guidance when assertions fail. | 4 / 5 |
Progressive Disclosure | Well-sectioned overview that keeps the main body to setup/running/CI, with the one deep-dive (`references/budgets.md` — a real file) held out one level deep and clearly signaled by a dedicated section stating what it contains and when to go there ("Start there when the project has no budgets yet"). | 5 / 5 |
Total | 17 / 20 Passed |