Install any skill in seconds. Free to start, no credit card required.
Get Started Free →基於 Core Web Vitals 的效能分析和 UX 評分。「分析效能」「改善速度」「檢查 Core Web Vitals」「頁面速度」等觸發。
.claude/skills/wasabeef-core-web-vitals-ux-d8ef41/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | 155% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 59% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 107% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 72% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 60% | 0% |
從使用者體驗的角度分析應用程式效能,並量化透過最佳化實現的感知速度提升。基於 Core Web Vitals 計算 UX 分數,並提出有優先序的最佳化策略。
text使用者體驗分數:B+ (78/100) ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ ⏱️ Core Web Vitals ├─ LCP (載入):2.3 秒 [良好] 目標<2.5 秒 ✅ ├─ INP (回應):95ms [良好] 目標<200ms ✅ ├─ CLS (視覺穩定):0.08 [良好] 目標<0.1 ✅ ├─ FCP (首次繪製):1.8 秒 [良好] 目標<1.8 秒 ✅ ├─ TTFB (伺服器):450ms [需要改善] 目標<200ms ⚠️ └─ TTI (可互動):3.5 秒 [需要改善] 目標<3.8 秒 ⚠️ 📊 使用者感知速度 ├─ 初始顯示:2.3 秒 [業界平均:3.0 秒] ├─ 頁面切換:1.1 秒 [業界平均:1.5 秒] ├─ 搜尋結果顯示:0.8 秒 [業界平均:1.2 秒] ├─ 表單提交:1.5 秒 [業界平均:2.0 秒] └─ 圖片載入:已實作延遲載入 ✅ 😊 使用者滿意度預測 ├─ 跳出率預測:12%(業界平均:20%) ├─ 完成率預測:78%(目標:85%) ├─ 推薦 NPS:+24(業界平均:+15) └─ 回訪率:65%(目標:70%) 📊 對使用者體驗的影響 ├─ 縮短顯示時間 0.5 秒 → 跳出率 -7% ├─ 降低跳出率 5% → 工作階段長度 +15% ├─ 改善搜尋 → 停留時間 +15% └─ 整體 UX 改善度:+25% 🎯 預期改善效果 (優先序排序) ├─ [P0] TTFB 改善 (導入 CDN)→ LCP -0.3 秒 = 感知速度 +15% ├─ [P1] JS 套件最佳化 → TTI -0.8 秒 = 互動時間 -20% ├─ [P2] 圖片最佳化 (WebP)→ 傳輸量 -40% = 載入時間 -25% └─ [P3] 快取策略 → 重複訪問時快 50%
bash# UX 分數的綜合分析 find . -name "*.js" -o -name "*.ts" | xargs wc -l | sort -rn | head -10 「計算 UX 效能分數並評估 Core Web Vitals」 # 效能瓶頸檢測 grep -r "for.*await\|forEach.*await" . --include="*.js" 「檢測非同步處理的瓶頸並分析對使用者感受的影響」 # 對使用者體驗的影響分析 grep -r "addEventListener\|setInterval" . --include="*.js" | grep -v "removeEventListener\|clearInterval" 「分析效能問題對使用者體驗的影響」
bash# 套件大小與載入時間 npm ls --depth=0 && find ./public -name "*.js" -o -name "*.css" | xargs ls -lh "識別套件大小與資源最佳化的改善點" # 資料庫效能 grep -r "SELECT\|findAll\|query" . --include="*.js" | head -20 "分析資料庫查詢的最佳化點" # 相依性的效能影響 npm outdated && npm audit "評估舊版相依性對效能的影響"
text改善 ROI = (時間縮短效果 + 品質提升)÷ 實作工時
| 優先度 | 使用者體驗提升 | 實作難度 | 時間縮短效果 | 具體例子 | 工時 | 效果 | | --------------------- | -------------- | -------- | ------------ | ---------- | ---- | --------- | | P0] 立即實作 | 高 | 低 | > 50% | 導入 CDN | 8h | 回應 -60% | | P1] 早期實作建議 | 高 | 中 | 20-50% | 圖片最佳化 | 16h | 載入 -30% | | P2] 計劃性實作 | 低 | 高 | 10-20% | 程式碼分割 | 40h | 初始 -15% | | P3] 保留/觀察 | 低 | 低 | < 10% | 微調最佳化 | 20h | 部分 -5% |
| 指標 | 改善幅度 | 感知速度提升 | 使用者滿意度 | 實作工時 | | ----------------- | -------- | ------------ | ------------- | -------- | | LCP(載入) | -0.5 秒 | +30% | 跳出率 -7% | 16h | | INP(回應) | -50ms | +15% | 壓力 -20% | 8h | | CLS(視覺穩定) | -0.05 | +10% | 誤操作 -50% | 4h | | TTFB(伺服器) | -200ms | +25% | 感知速度 +40% | 24h | | TTI(可互動) | -1.0 秒 | +35% | 完成率 +15% | 32h | | 套件大小 | -30% | +20% | 首次訪問 +25% | 16h |
bash# 效能分析 node --prof app.js clinic doctor -- node app.js # 套件分析 npx webpack-bundle-analyzer lighthouse --chrome-flags="--headless"
sql-- 查詢分析 EXPLAIN ANALYZE SELECT ... SHOW SLOW LOG;
bash# React 效能 grep -r "useMemo\|useCallback" . --include="*.jsx" # 資源分析 find ./src -name "*.png" -o -name "*.jpg" | xargs ls -lh
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-06 | fail→pass | 10,254 | 11,488 | +12% | 1 | 1 | 0% | 1,544 | 3,938 | +155% | 0 | 0 | — |
case-01 | fail→pass | 23,859 | 24,909 | +4% | 1 | 1 | 0% | 3,842 | 6,114 | +59% | 0 | 0 | — |
case-02 | fail→pass | 18,863 | 23,496 | +25% | 1 | 1 | 0% | 2,940 | 6,075 | +107% | 0 | 0 | — |
case-03 | fail→fail | 22,080 | 20,604 | -7% | 1 | 1 | 0% | 3,436 | 5,240 | +53% | 0 | 0 | — |
case-04 | fail→fail | 12,545 | 13,168 | +5% | 1 | 1 | 0% | 2,125 | 4,175 | +96% | 0 | 0 | — |
case-05 | pass→pass | 12,142 | 6,580 | -46% | 1 | 1 | 0% | 1,915 | 3,052 | +59% | 0 | 0 | — |
case-16 | pass→pass | 9,328 | 6,994 | -25% | 1 | 1 | 0% | 1,692 | 3,191 | +89% | 0 | 0 | — |
case-07 | pass→pass | 5,630 | 4,250 | -25% | 1 | 1 | 0% | 944 | 2,720 | +188% | 0 | 0 | — |
case-08 | fail→pass | 12,599 | 9,330 | -26% | 1 | 1 | 0% | 2,024 | 3,480 | +72% | 0 | 0 | — |
case-09 | fail→pass | 17,988 | 14,293 | -21% | 1 | 1 | 0% | 2,738 | 4,377 | +60% | 0 | 0 | — |
case-10 | pass→pass | 12,606 | 5,804 | -54% | 1 | 1 | 0% | 1,967 | 2,969 | +51% | 0 | 0 | — |
case-22 | pass→fail | 12,426 | 14,323 | +15% | 1 | 1 | 0% | 2,038 | 4,470 | +119% | 0 | 0 | — |
case-11 | fail→pass | 9,561 | 3,449 | -64% | 1 | 1 | 0% | 1,372 | 2,645 | +93% | 0 | 0 | — |
case-12 | fail→fail | 12,617 | 11,659 | -8% | 1 | 1 | 0% | 2,091 | 3,940 | +88% | 0 | 0 | — |
case-13 | pass→pass | 16,543 | 24,031 | +45% | 1 | 1 | 0% | 2,840 | 6,203 | +118% | 0 | 0 | — |
case-14 | fail→fail | 10,570 | 11,346 | +7% | 1 | 1 | 0% | 1,834 | 3,966 | +116% | 0 | 0 | — |
case-15 | fail→fail | 15,786 | 16,551 | +5% | 1 | 1 | 0% | 2,871 | 4,993 | +74% | 0 | 0 | — |
case-17 | fail→pass | 8,190 | 4,458 | -46% | 1 | 1 | 0% | 1,419 | 2,748 | +94% | 0 | 0 | — |
case-18 | fail→pass | 10,781 | 2,621 | -76% | 1 | 1 | 0% | 1,693 | 2,484 | +47% | 0 | 0 | — |
case-19 | fail→pass | 15,250 | 7,399 | -51% | 1 | 1 | 0% | 2,673 | 3,385 | +27% | 0 | 0 | — |
case-20 | pass→pass | 18,004 | 14,949 | -17% | 1 | 1 | 0% | 2,958 | 4,362 | +47% | 0 | 0 | — |
case-21 | fail→fail | 15,496 | 14,652 | -5% | 1 | 1 | 0% | 2,743 | 4,397 | +60% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +36 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.