Content
57%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.
The body is a clean, well-organized overview that correctly pushes code examples and framework detail into references/rule.md, giving excellent progressive disclosure. Its weaknesses are duplicated statistics across sections, Check/Fix sections that name no measurement tool or command, and a missing re-measure step after applying fixes.
Suggestions
Make the Check section executable: name the measurement tool and pass criteria, e.g. 'Run Lighthouse (or Chrome DevTools Performance tab) with Slow 4G throttling; pass if LCP < 2.5s and full load < 3s' — or at minimum name Lighthouse/DevTools as the method rather than an unadorned 'measure'.
Add a re-verification step after Fix: 'After applying optimizations, re-run the same measurement to confirm load time is under 3 seconds' to close the measure → optimize → re-measure loop.
De-duplicate the 53% statistic (stated in both the intro and Quick Reference) and let the 'Explain' section reference the Quick Reference facts instead of restating bounce-rate impact.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The 53% abandonment statistic is stated twice ('53% of mobile users abandon sites that take longer than 3 seconds' in the intro and '53% of mobile users abandon sites taking over 3 seconds' in Quick Reference), and the bounce-rate theme recurs across the intro, Quick Reference, and the 'Explain' section. Mostly efficient but with noticeable duplication that could be tightened — anchor 3, not 4, since the redundancy goes beyond minor trimming. | 3 / 5 |
Actionability | 'Measure the page load time and verify it's under 3 seconds' and 'Optimize page load time through lazy loading, CDN usage, caching strategies' give high-level direction with no tool, command, or code in the body. Concrete specifics do appear ('Core Web Vitals: LCP, FID/INP, CLS', 'Test on throttled 3G') and the pointer to references/rule.md leads to complete executable examples, but key details — which tool, how to measure — are missing. Anchor 3, not 4, because the core Check/Fix actions lack executable guidance in the body. | 3 / 5 |
Workflow Clarity | The Check → Fix → Explain → Code Review sequence is clearly sectioned and logically ordered, and the check step validates ('verify it's under 3 seconds on a standard connection'). However, there is no re-verification after applying fixes — no measure → optimize → re-measure feedback loop — so checkpoints are incomplete. Anchor 3, not 4: the post-fix checkpoint is absent entirely, not just minor. | 3 / 5 |
Progressive Disclosure | The body is a ~30-line well-sectioned overview, with implementation details, code examples, and framework guidance correctly split into references/rule.md — one level deep, clearly signaled ('For full implementation details, code examples, and framework-specific guidance, see references/rule.md'), and that file exists and does not nest further references. This matches the anchor 5 pattern of a clear overview with well-signaled one-level-deep references. | 5 / 5 |
Total | 14 / 20 Passed |