Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →テクニカル SEO、オンページ最適化、構造化データ、Core Web Vitals、およびコンテンツ戦略にわたる SEO 改善の監査、計画、実施。ユーザーが検索可視性の向上、SEO 修正、スキーママークアップ、サイトマップ/robots の作業、またはキーワードマッピングを希望する場合に使用します。
.claude/skills/affaan-m-seo/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -7% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 30% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-03 | ✓→✗ | ▼ Worse | 101% | 0% |
| case-22 | ✓→✗ | ▼ Worse | 81% | 0% |
通过技术正确性、性能和内容相关性提升搜索可见性，而非依赖花哨手段。
在以下情况使用此技能：
robots.txt 应允许重要页面并屏蔽低价值内容noindexArticle / BlogPostingProduct 和 OfferBreadcrumbListFAQPageH1H2 和 H3 应反映实际内容层级text主要主题 - 特定修饰词 | 品牌
text行动 + 主题 + 价值主张 + 一个支撑细节
json{ "@context": "https://schema.org", "@type": "Article", "headline": "Page Title Here", "author": { "@type": "Person", "name": "Author Name" }, "publisher": { "@type": "Organization", "name": "Brand Name" } }
text[HIGH] 产品页面上的重复标题标签 位置：src/routes/products/[slug].tsx 问题：动态标题会折叠为相同的默认字符串，这会削弱相关性并产生重复信号。 修复：使用产品名称和主要类别为每个产品生成唯一的标题。
| 反模式 | 修复方法 | | --- | --- | | 关键词堆砌 | 优先为用户写作 | | 内容单薄的近似重复页面 | 合并或差异化处理 | | 为不存在的内容添加架构 | 使架构与实际内容匹配 | | 未检查实际页面就提供内容建议 | 先阅读真实页面 | | 泛泛的“改进SEO”输出 | 将每条建议与具体页面或资源关联 |
seo-specialistfrontend-patternsbrand-voicemarket-research| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-10 | pass→pass | 8,201 | 8,850 | +8% | 1 | 1 | 0% | 1,402 | 2,457 | +75% | 0 | 0 | — |
case-01 | fail→pass | 30,186 | 21,889 | -27% | 1 | 1 | 0% | 5,355 | 4,962 | -7% | 0 | 0 | — |
case-02 | pass→pass | 8,809 | 6,842 | -22% | 1 | 1 | 0% | 1,384 | 2,144 | +55% | 0 | 0 | — |
case-03 | pass→fail | 7,325 | 7,743 | +6% | 1 | 1 | 0% | 1,148 | 2,308 | +101% | 0 | 0 | — |
case-04 | fail→pass | 10,889 | 8,806 | -19% | 1 | 1 | 0% | 1,944 | 2,519 | +30% | 0 | 0 | — |
case-05 | pass→pass | 14,542 | 9,554 | -34% | 1 | 1 | 0% | 2,726 | 2,869 | +5% | 0 | 0 | — |
case-06 | pass→pass | 11,352 | 7,886 | -31% | 1 | 1 | 0% | 2,028 | 2,311 | +14% | 0 | 0 | — |
case-07 | pass→pass | 14,151 | 7,341 | -48% | 1 | 1 | 0% | 2,672 | 2,315 | -13% | 0 | 0 | — |
case-08 | pass→pass | 10,139 | 9,933 | -2% | 1 | 1 | 0% | 1,611 | 2,500 | +55% | 0 | 0 | — |
case-09 | pass→pass | 11,695 | 9,745 | -17% | 1 | 1 | 0% | 1,859 | 2,594 | +40% | 0 | 0 | — |
case-11 | pass→pass | 10,131 | 7,004 | -31% | 1 | 1 | 0% | 1,518 | 2,228 | +47% | 0 | 0 | — |
case-12 | pass→pass | 15,334 | 12,576 | -18% | 1 | 1 | 0% | 2,353 | 2,995 | +27% | 0 | 0 | — |
case-13 | pass→pass | 8,776 | 7,889 | -10% | 1 | 1 | 0% | 1,341 | 2,232 | +66% | 0 | 0 | — |
case-14 | pass→pass | 12,270 | 9,268 | -24% | 1 | 1 | 0% | 1,935 | 2,509 | +30% | 0 | 0 | — |
case-15 | pass→pass | 6,280 | 6,167 | -2% | 1 | 1 | 0% | 1,216 | 2,056 | +69% | 0 | 0 | — |
case-16 | pass→pass | 13,542 | 11,477 | -15% | 1 | 1 | 0% | 2,076 | 2,672 | +29% | 0 | 0 | — |
case-17 | pass→pass | 12,373 | 11,871 | -4% | 1 | 1 | 0% | 1,881 | 2,821 | +50% | 0 | 0 | — |
case-18 | pass→pass | 14,526 | 11,877 | -18% | 1 | 1 | 0% | 2,686 | 3,203 | +19% | 0 | 0 | — |
case-19 | pass→pass | 12,812 | 9,379 | -27% | 1 | 1 | 0% | 1,953 | 2,537 | +30% | 0 | 0 | — |
case-20 | fail→pass | 22,848 | 13,739 | -40% | 1 | 1 | 0% | 2,118 | 3,022 | +43% | 0 | 0 | — |
case-21 | pass→pass | 20,084 | 20,322 | +1% | 1 | 1 | 0% | 2,991 | 4,164 | +39% | 0 | 0 | — |
case-22 | pass→fail | 14,075 | 19,112 | +36% | 1 | 1 | 0% | 1,927 | 3,488 | +81% | 0 | 0 | — |
case-23 | fail→fail | 6,671 | 7,569 | +13% | 1 | 1 | 0% | 1,276 | 2,505 | +96% | 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. 23 cases were attempted. The headline lift of +4 percentage points is the difference between those two pass rates over the 23 comparable cases. 2 cases got worse with the skill loaded, and they are 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.