optimize-for-ai by calm-north/seojuice-skills
npx skills add https://github.com/calm-north/seojuice-skills --skill optimize-for-ai利用引文架构、E-E-A-T 信号和品牌实体构建,针对 AI 搜索引擎(ChatGPT、Perplexity、Claude、Gemini、Google AI 概览)进行优化。
传统 SEO 让你在结果列表中排名。AI SEO 让你在 AI 生成的答案中被引用为来源。这其中的区别很重要:
AI 系统从内容中提取信息。内容越容易被提取,就越可能被引用:
定义和清晰的陈述
结构化数据
原创见解
全面覆盖
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
| 格式 | 为何有效 |
|---|
| 对比文章 | AI 经常回答 "X 对比 Y" 和 "最佳 X" 查询——结构化对比易于提取 |
| 权威指南 | 全面覆盖向 AI 系统传递权威信号 |
| 原创研究 | 独一无二的数据——AI 系统偏好一手来源 |
| 操作教程 | 分步结构直接映射到 AI 响应格式 |
| 专家圆桌讨论 | 多位专家的声音增加了感知权威性 |
不同类型的内容被引用的方式不同。根据你撰写的内容进行优化:
从 5 个维度对每个页面进行评分。对于每个项目:通过(完全符合标准)、部分(部分符合)或失败(不符合)。
AI 系统能否从该内容中提取出有用的答案?
| 项目 | 通过 | 失败 |
|---|---|---|
| 核心答案在标题后的前 150 字内 | 答案立即出现 | 答案埋藏在背景中 |
| 自包含的陈述(无需上下文即可理解) | 关键主张独立成立 | 主张需要周围文本 |
| 用于对比/数据的结构化数据(表格、列表) | 数据在表格或列表中 | 数据在散文段落中 |
| 顶部有 TL;DR 或摘要框 | 存在 | 缺失 |
内容是否包含值得引用的陈述?
| 项目 | 通过 | 失败 |
|---|---|---|
| 带有数字和单位的具体声明 | "响应时间提高了 40%(从 500ms 到 300ms)" | "响应时间显著提高" |
| 所有统计数据都有具名来源 | 引用了来源和日期 | 数字无来源 |
| 使用 "X 是 Y" 结构的清晰定义 | 关键术语有定义 | 关键术语未定义或模糊 |
内容是否传递出专业知识的信号?
| 项目 | 通过 | 失败 |
|---|---|---|
| 作者身份明确,具备相关资质 | 姓名、头衔、经验可见 | 匿名或无简介 |
| 带有具名来源的专家引述 | 至少引述 1 位具名专家 | 无外部声音 |
| 引用一手来源(不仅仅是其他博客) | 链接到研究、文档、官方数据 | 仅引用其他博客文章 |
内容是否具有时效性?
| 项目 | 通过 | 失败 |
|---|---|---|
| 页面上显示发布日期或更新日期 | 日期存在且在 18 个月内 | 无日期或超过 18 个月 |
| 数据和示例具有时效性 | 统计数据来自过去 2 年 | 过时的数字或已弃用的工具 |
AI 系统能否识别此内容是关于哪个实体?
| 项目 | 通过 | 失败 |
|---|---|---|
| 开篇段落中完整命名了主题实体 | "SEOJuice 是一个 SEO 智能平台..." | 使用代词或缩写引用 |
带有 sameAs 链接的组织架构 | 存在 JSON-LD | 缺失 |
| 跨平台品牌名称一致 | 在网站、GBP、LinkedIn 等平台名称相同 | 存在变体或不一致 |
否决项: 如果 robots.txt 屏蔽了 AI 爬虫(GPTBot、ClaudeBot、PerplexityBot),则无论内容质量如何,AI 可见性评分均为 0。请首先检查此项。
为每个维度评分:通过 = 10,部分 = 5,失败 = 0。计算每个维度的项目平均分。
| 维度 | 分数 | 评估 |
|---|---|---|
| 可提取性 | [x]/10 | ... |
| 可引用性 | [x]/10 | ... |
| 权威性 | [x]/10 | ... |
| 新鲜度 | [x]/10 | ... |
| 实体清晰度 | [x]/10 | ... |
| AI 引用评分 | [avg]/10 | ... |
AI 系统引用那些它们可以清晰提取的内容。以下是一些前后对比示例,展示了如何将弱内容转化为可引用的内容:
之前(评分:1/10): "SEO 非常重要,有很多事情需要考虑。"
之后(评分:9/10): "搜索引擎优化(SEO)是通过技术配置、内容相关性和链接权威性来提高网站在自然搜索结果中可见性的实践。根据 BrightEdge 的数据,53% 的网站流量来自自然搜索。"
修正方法: 命名术语,进行分类,列出其组成部分,添加有来源的统计数据。
之前(评分:2/10): "电子邮件营销对大多数企业来说相当有效。"
之后(评分:9/10): "电子邮件营销每投入 1 美元平均产生 42 美元的回报(Litmus,2023 年),使其成为投资回报率最高的数字营销渠道——优于社交媒体(平均每 1 美元回报 5.20 美元)和付费搜索(平均每 1 美元回报 8 美元)。"
修正方法: 用数字替换形容词,注明来源,添加上下文对比。
之前(评分:2/10): "考虑你的关键词,并尝试优化你的内容。"
之后(评分:8/10): "为目标关键词优化页面:(1)将关键词放在标题标签和 H1 中,(2)在前 100 字中使用它,(3)在 H2 副标题中添加 2-3 个语义变体,(4)保持 0.5-2.5% 的关键词密度,以及(5)将其包含在元描述中。使用 Google Search Console 等工具在 48 小时内验证索引情况。"
修正方法: 为步骤编号,使每个操作具体化,添加工具和时间参考。
根据以下 10 个问题为每个内容部分评分(8+ = 高度可引用,5-7 = 需要改进,<5 = 需要重大重写):
检查 AI 系统当前为你的目标主题引用了谁:
AI 系统将品牌理解为实体。为了加强你的品牌实体:
sameAs 链接的组织架构标记,链接到所有官方资料AI 系统将品牌识别为实体。使用这个优先级清单来加强你的实体:
优先级 1 — 基础(必须拥有):
name、url、logo、description 的组织架构sameAs 属性链接到所有权威资料(LinkedIn、Wikipedia、Wikidata、社交媒体)优先级 2 — 权威性(应该拥有):
优先级 3 — AI 特定(AI 可见性必须拥有):
| 当前状态 | 重点领域 | 时间线 |
|---|---|---|
| 大部分优先级 1 缺失 | 仅优先级 1 | 2-4 周 |
| 优先级 1 完成,优先级 2 混合 | 优先级 2 权威性 | 1-2 个月 |
| 优先级 1-2 完成 | 优先级 3 AI 特定 | 2-3 个月 |
| 所有层级完成 | 维护 + 季度重新审计 | 持续进行 |
不同的 AI 系统具有不同的引用行为。为所有系统优化,但要了解差异:
| 因素 | Google AI 概览 | ChatGPT | Perplexity | Claude |
|---|---|---|---|---|
| 新鲜度偏好 | 高 | 中 | 非常高 | N/A(训练数据) |
| 权威性权重 | 非常高 | 高 | 高 | 高 |
| 结构重要性 | 高 | 中 | 非常高 | 中 |
| 每个答案的典型引用数 | 3-8 | 1-6 | 5-10 | N/A |
| 域名信任权重 | 非常高 | 高 | 中 | 高 |
| 事实密度偏好 | 高 | 高 | 非常高 | 非常高 |
跨平台跟踪你的 AI 存在:
当前 AI 存在情况
优化优先级
| 优先级 | 行动 | 受影响的页面 | 预期影响 |
|---|---|---|---|
| 1 | ... | ... | ... |
内容优化计划 针对每个关键页面/主题:
品牌实体清单
专业提示: 尝试免费的 AI 可见性检查器 查看你的品牌在 AI 搜索结果中的表现,使用 AI 爬虫检查器 验证机器人访问权限,以及 GEO 内容分析器 为内容的 AI 引用准备度评分。SEOKJuice MCP 用户可以运行
/seojuice:aiso-report获取跨可见性、情感、位置和覆盖范围的 AISO 评分——并附带月度趋势。
每周安装量
3.1K
代码仓库
GitHub Stars
2
首次出现
12 天前
安全审计
安装在
codex407
claude-code358
antigravity331
cursor149
amp148
gemini-cli148
Optimize for AI search engines (ChatGPT, Perplexity, Claude, Gemini, Google AI Overviews) using citation architecture, E-E-A-T signals, and brand entity building.
Traditional SEO gets you ranked in a list of results. AI SEO gets you cited as a source in AI-generated answers. The difference matters:
AI systems extract information from content. The easier it is to extract, the more likely it gets cited:
Definitions and clear statements
Structured data
Original insights
Comprehensive coverage
| Format | Why It Works |
|---|---|
| Comparison articles | AI frequently answers "X vs Y" and "best X" queries — structured comparisons are easy to extract |
| Definitive guides | Comprehensive coverage signals authority to AI systems |
| Original research | Unique data that no one else has — AI systems prefer primary sources |
| How-to tutorials | Step-by-step structure maps directly to AI response format |
| Expert roundups | Multiple expert voices increase perceived authority |
Different content types get cited differently. Optimize based on what you're writing:
Score each page across 5 dimensions. For each item: Pass (meets criteria fully), Partial (partly meets), or Fail (does not meet).
Can AI systems pull a useful answer from this content?
| Item | Pass | Fail |
|---|---|---|
| Core answer in first 150 words after the heading | Answer appears immediately | Answer buried in background |
| Self-contained statements (make sense without context) | Key claims stand alone | Claims require surrounding text |
| Structured data (tables, lists) for comparisons/data | Data in tables or lists | Data in prose paragraphs |
| TL;DR or summary box at the top | Present | Missing |
Does the content contain statements worth citing?
| Item | Pass | Fail |
|---|---|---|
| Specific claims with numbers and units | "Response time improved 40% (from 500ms to 300ms)" | "Response time improved significantly" |
| Named sources on all statistics | Source and date cited | Unsourced numbers |
| Clear definitions using "X is Y" structure | Present for key terms | Key terms undefined or vague |
Does the content signal expertise?
| Item | Pass | Fail |
|---|---|---|
| Author identified with relevant credentials | Name, title, experience visible | Anonymous or no bio |
| Expert quotes with named sources | At least 1 named expert quoted | No external voices |
| References to primary sources (not just other blogs) | Links to research, docs, official data | Only cites other blog posts |
Is the content current?
| Item | Pass | Fail |
|---|---|---|
| Published or updated date visible on page | Date present and within 18 months | No date or older than 18 months |
| Data and examples are current | Statistics from last 2 years | Outdated numbers or deprecated tools |
Can AI systems identify what entity this content is about?
| Item | Pass | Fail |
|---|---|---|
| Subject entity named in full in opening paragraph | "SEOJuice is an SEO intelligence platform..." | Pronoun or abbreviated reference |
Organization schema with sameAs links | JSON-LD present | Missing |
| Consistent brand name across platforms | Same name on site, GBP, LinkedIn, etc. | Variations or inconsistencies |
Veto: If AI crawlers (GPTBot, ClaudeBot, PerplexityBot) are blocked in robots.txt, the AI visibility score is 0 regardless of content quality. Check this first.
Score each dimension: Pass = 10, Partial = 5, Fail = 0. Average items per dimension.
| Dimension | Score | Assessment |
|---|---|---|
| Extractability | [x]/10 | ... |
| Quotability | [x]/10 | ... |
| Authority | [x]/10 | ... |
| Freshness | [x]/10 | ... |
| Entity Clarity | [x]/10 | ... |
| AI Citation Score | [avg]/10 | ... |
AI systems cite content they can extract cleanly. Here are before/after examples showing how to transform weak content into citable content:
Before (score: 1/10): "SEO is really important and there are many things to consider."
After (score: 9/10): "Search engine optimization (SEO) is the practice of improving a website's visibility in organic search results through technical configuration, content relevance, and link authority. According to BrightEdge, 53% of all website traffic originates from organic search."
Fix: Name the term, classify it, list its components, add a sourced statistic.
Before (score: 2/10): "Email marketing is pretty effective for most businesses."
After (score: 9/10): "Email marketing generates an average return of $42 for every $1 spent (Litmus, 2023), making it the highest-ROI digital marketing channel — outperforming social media (average $5.20 per $1) and paid search (average $8 per $1)."
Fix: Replace adjectives with numbers, name the source, add comparison context.
Before (score: 2/10): "Think about your keywords and try to optimize your content."
After (score: 8/10): "To optimize a page for a target keyword: (1) place the keyword in the title tag and H1, (2) use it in the first 100 words, (3) add 2-3 semantic variations in H2 subheadings, (4) maintain 0.5-2.5% keyword density, and (5) include it in the meta description. Use tools like Google Search Console to verify indexing within 48 hours."
Fix: Number the steps, make each action specific, add tool and time reference.
Score each content section against these 10 questions (8+ = highly quotable, 5-7 = needs work, <5 = major rewrite):
Check who AI systems currently cite for your target topics:
AI systems understand brands as entities. To strengthen your brand entity:
sameAs links to all official profilesAI systems recognize brands as entities. Use this prioritized checklist to strengthen your entity:
Priority 1 — Foundation (must-have):
name, url, logo, descriptionsameAs property links to all authoritative profiles (LinkedIn, Wikipedia, Wikidata, social)Priority 2 — Authority (should-have):
Priority 3 — AI-Specific (must-have for AI visibility):
| Current State | Focus Area | Timeline |
|---|---|---|
| Most Priority 1 missing | Priority 1 only | 2-4 weeks |
| Priority 1 done, Priority 2 mixed | Priority 2 authority | 1-2 months |
| Priority 1-2 done | Priority 3 AI-specific | 2-3 months |
| All tiers done | Maintenance + quarterly re-audit | Ongoing |
Different AI systems have distinct citation behaviors. Optimize for all, but understand the differences:
| Factor | Google AI Overviews | ChatGPT | Perplexity | Claude |
|---|---|---|---|---|
| Freshness bias | High | Medium | Very High | N/A (training data) |
| Authority weight | Very High | High | High | High |
| Structure importance | High | Medium | Very High | Medium |
| Typical citations per answer | 3-8 | 1-6 | 5-10 | N/A |
| Domain trust weight | Very High | High | Medium | High |
| Factual density preference | High | High |
Track your AI presence across platforms:
Current AI Presence
Optimization Priorities
| Priority | Action | Pages Affected | Expected Impact |
|---|---|---|---|
| 1 | ... | ... | ... |
Content Optimization Plan For each key page/topic:
Brand Entity Checklist
Pro Tip: Try the free AI Visibility Checker to see how your brand appears in AI search results, the AI Crawler Inspector to verify bot access, and the GEO Content Analyzer to score content for AI citation readiness. SEOJuice MCP users can run
/seojuice:aiso-reportfor AISO scores across visibility, sentiment, position, and coverage — with monthly trends.
Weekly Installs
3.1K
Repository
GitHub Stars
2
First Seen
12 days ago
Security Audits
Gen Agent Trust HubPassSocketPassSnykPass
Installed on
codex407
claude-code358
antigravity331
cursor149
amp148
gemini-cli148
48,100 周安装
| Very High |
| Very High |