llm-optimized-content
关于
This Claude Skill optimizes content for AI search engines like ChatGPT and Perplexity using Generative Engine Optimization (GEO) principles. It ensures AI models understand, deem reliable, and directly cite your content in their responses. Use it when traditional SEO is insufficient and you need to prepare for AI-powered "zero-click" search results.
快速安装
Claude Code
推荐npx skills add guia-matthieu/clawfu-skills -a claude-code/plugin add https://github.com/guia-matthieu/clawfu-skillsgit clone https://github.com/guia-matthieu/clawfu-skills.git ~/.claude/skills/llm-optimized-content在 Claude Code 中复制并粘贴此命令以安装该技能
技能文档
LLM-Optimized Content & GEO
Create content optimized for AI-powered search using Generative Engine Optimization (GEO) principles.
Purpose
Create content optimized for both traditional search engines AND AI/LLM-powered search (ChatGPT, Perplexity, Google AI Overviews, Claude) using Generative Engine Optimization (GEO) principles.
What is GEO?
Generative Engine Optimization = Practices to ensure AI models:
- UNDERSTAND your content
- Judge it RELEVANT and RELIABLE
- USE it to formulate direct responses
Strategic Importance: Gartner predicts 25% drop in traditional search volume by 2026. Queries now average 23 words (vs. 4 before).
SEO vs GEO: Critical Difference
| Criteria | SEO (Traditional) | GEO (New Frontier) |
|---|---|---|
| Objective | Rank page on Google SERP | Become the SOURCE in AI response |
| Metric | Position on results page | Mention/citation in generated response |
| Mechanic | Links, backlinks, domain authority | Language, semantics, contextual relevance |
| User Result | Click to your site | Info consumed without click |
| Optimization | Keywords, tags, structure | Semantic clarity, E-E-A-T, direct answers |
GEO Optimization Strategies
1. Semantic Clarity & Relevance
Principle: AI must understand CONTEXT, not just keywords.
Actions:
- Dense in MEANING, not just keywords
- Natural vocabulary, no keyword stuffing
- Clear context in first 1-2 sentences
- Define technical terms at first appearance
❌ "Our SEO GEO optimization ranking solution"
✅ "GEO (Generative Engine Optimization) optimizes content so AI like
ChatGPT cites your brand in their responses."
2. Direct Answers (Answer-First)
Principle: Provide key answer at the BEGINNING, not the end.
Actions:
- Main answer in 1st paragraph
- Format: "Question → Immediate answer → Development"
- No "teasing" (saving answer for later)
Optimal Structure:
## [Question]
[Direct answer in 1-2 sentences]
[Detailed development]
[Concrete examples]
[Conclusion/Action]
Example:
## What is ethical copywriting?
Ethical copywriting is a persuasive writing approach that excludes
manipulation and is founded on transparency, honesty, and respect
for the prospect.
[Details follow...]
3. AI-Parser Optimized Structure
Principle: AI favors content that's easy to analyze ("parse").
Headers (H1-H6)
- Descriptive and informative (not clickbait)
- Strict logical hierarchy
- Contain keywords naturally
✅ "## How to create a landing page that converts?"
❌ "## The secret marketers don't want you to know"
Bullet Lists
- For enumerations, steps, comparisons
- Preferred over long paragraphs
- Start each item with strong keyword
Tables
- For comparisons, specifications, data
- Clear, descriptive headers
- Concise cells
Question-Answer Format (FAQ)
- H3 structure for questions
- Immediate answer after
- Ideal for featured snippets AND GEO
4. E-E-A-T: Credibility for AI
E-E-A-T = Expertise + Experience + Authority + Trustworthiness
Expertise
- Demonstrate deep knowledge
- Use precise domain terminology
- Reference research/studies
Experience
- First-hand concrete examples
- Real case studies
- Measurable results obtained
Authority
- Author bio with credentials
- Citations by recognized experts
- Publications in reference media
- Certifications, awards
Trustworthiness
- Cited and dated sources
- Transparency about limits/biases
- Accessible contact and legal notices
- HTTPS, professional design
Concrete Signals to Integrate:
"According to a [University X] study published in [year]..."
"Based on our analysis of [X] clients over [period]..."
"Source: [URL of official document]"
"Last updated: [date]"
5. Optimal Length & Depth
Data:
- Articles 2000-3000 words = sweet spot for GEO
- But: Quality > Quantity
- Depth on ONE subject > Overview of several
Ideal GEO Article Structure:
1. Intro with direct answer (100-150 words)
2. Definition/Context (200-300 words)
3. In-depth development (1500-2000 words)
- Subsections with H2/H3
- Concrete examples
- Tables/lists
4. Case studies/Proof (300-500 words)
5. Actionable conclusion (100-150 words)
6. Freshness & Updates
Principle: AI favors RECENT information.
Actions:
- Explicitly date content
- Mention "updated on [date]"
- Recent statistics (< 2 years)
- References to relevant current events
7. Citation-Friendly Format
Principle: Facilitate extraction and citation by AI.
Techniques:
- Autonomous sentences (understandable alone)
- Statistics with immediate context
- Clear, concise definitions
❌ "That's 37% more."
✅ "Ethical copywriting increases conversion rates by 37% on average
compared to manipulative approaches."
GEO by Content Type
Blog Articles
- Title = Natural question
- Answer in meta description AND 1st paragraph
- H2 subtitles = Secondary questions
- FAQ at end of article
Product/Service Pages
- Main benefit in H1 (visualizable + falsifiable)
- Technical specs in table
- Concrete use cases (storytelling)
- Testimonials with context
Landing Pages
- Clear value proposition in hero
- Sections with descriptive H2s
- Measurable social proof
- FAQ for common objections
Content Structure Template
# [H1: Clear Title with Primary Topic]
[Opening paragraph with direct answer to main query]
## Quick Summary
- **What:** [One-line definition]
- **Why it matters:** [Key benefit/importance]
- **Key takeaway:** [Main action or insight]
## Table of Contents
[For posts >1500 words]
## [H2: First Major Section]
[Direct answer/definition first, then elaboration]
### [H3: Supporting Detail]
[Expand with examples, data, explanations]
## [H2: How to [Action]]
1. **Step 1:** [Clear instruction]
- [Supporting detail]
2. **Step 2:** [Clear instruction]
- [Supporting detail]
## [H2: Comparison Section] (if applicable)
| Aspect | Option A | Option B |
|--------|----------|----------|
| [Criteria] | [Value] | [Value] |
## Key Takeaways
- [Bullet summary 1]
- [Bullet summary 2]
- [Bullet summary 3]
## Frequently Asked Questions
### [Natural language question]?
[2-3 sentence direct answer]
### [Natural language question]?
[2-3 sentence direct answer]
## Conclusion
[Summary + clear call to action]
---
*Last updated: [Date]*
*Sources: [Citations]*
Common Question Patterns to Answer
For any topic, consider addressing:
- What is [topic]? → Clear definition
- How does [topic] work? → Process explanation
- Why is [topic] important? → Benefits/significance
- How to [action with topic]? → Step-by-step guide
- [Topic] vs [alternative]? → Comparison
- Best [topic] for [use case]? → Recommendations
- How much does [topic] cost? → Pricing/investment
- How long does [topic] take? → Timeline expectations
GEO Checklist
Structure
- Descriptive title (not clickbait)
- Direct answer in first 150 words
- Informative H2/H3 headers
- Bullet lists for key points
- Table for comparisons/data
Content
- Dense in meaning (clear context)
- Autonomous citable sentences
- Statistics with context
- Definitions at first mention
- Concrete examples
Credibility (E-E-A-T)
- Cited and dated sources
- Author bio with expertise
- First-hand examples
- Visible publication/update date
Length
- 2000-3000 words for main subject
- Depth > Overview
- Each section developed
Technical
- Descriptive URL
- Meta description = direct answer
- Schema markup (if applicable)
- Images with descriptive alt text
Writing for 3 Audiences
Simultaneously optimize for:
- Human → Persuasion, engagement, emotional connection
- Ranking Algorithm (SEO) → Keywords, links, structure
- Generation Algorithm (GEO) → Semantic clarity, E-E-A-T, direct answers
Errors to Avoid
❌ Keyword stuffing → AI detects and devalues ❌ Clickbait headlines → AI seeks descriptive info ❌ Superficial content → AI prefers depth ❌ Absence of sources → Trustworthiness questioned ❌ Unexplained jargon → Missing context ❌ Buried answers → AI scans the beginning
What Claude Does vs What You Decide
| Claude handles | You provide |
|---|---|
| Applying GEO structure and formatting | Topic expertise and unique insights |
| Creating FAQ sections from content | Validation of accuracy |
| Optimizing for semantic clarity | E-E-A-T signals (credentials, experience) |
| Suggesting citable sentence structures | Source verification and dates |
| Generating comparison tables | Strategic decisions on positioning |
Skill Boundaries
This skill excels for:
- Blog articles and long-form content
- Product/service pages
- FAQ and knowledge base content
- Landing pages with informational elements
This skill is NOT ideal for:
- Pure conversion copy → Use copywriting skills
- Technical documentation → Structure matters more than GEO
- Internal communications → No AI discoverability needed
Iteration Guide
| Pass | Focus | Action |
|---|---|---|
| 1st | Structure | Apply GEO template, answer-first format |
| 2nd | E-E-A-T | Add citations, credentials, dates |
| 3rd | Citability | Make sentences autonomous and extractable |
| 4th | Checklist | Run GEO checklist to verify all elements |
Skill Metadata
name: llm-optimized-content
category: seo-tools
subcategory: geo
version: 2.0
author: GUIA
source_expert: GEO Framework (Gartner, industry research)
difficulty: intermediate
mode: cyborg
tags: [geo, seo, llm, ai-search, content, perplexity, chatgpt]
created: 2026-01-28
updated: 2026-02-03
GitHub 仓库
相关推荐技能
content-collections
元Content Collections 是一个 TypeScript 优先的构建工具,可将本地 Markdown/MDX 文件转换为类型安全的数据集合。它专为构建博客、文档站和内容密集型 Vite+React 应用而设计,提供基于 Zod 的自动模式验证。该工具涵盖从 Vite 插件配置、MDX 编译到生产环境部署的完整工作流。
polymarket
元这个Claude Skill为开发者提供完整的Polymarket预测市场开发支持,涵盖API调用、交易执行和市场数据分析。关键特性包括实时WebSocket数据流,可监控实时交易、订单和市场动态。开发者可用它构建预测市场应用、实施交易策略并集成实时市场预测功能。
creating-opencode-plugins
元该Skill帮助开发者创建OpenCode插件,用于接入命令、文件、LSP等25+种事件。它提供了插件结构、事件API规范和JavaScript/TypeScript实现模式,适合需要拦截操作、扩展功能或自定义事件处理的场景。开发者可通过它快速构建响应式模块来增强OpenCode AI助手的能力。
sglang
元SGLang是一个专为LLM设计的高性能推理框架,特别适用于需要结构化输出的场景。它通过RadixAttention前缀缓存技术,在处理JSON、正则表达式、工具调用等具有重复前缀的复杂工作流时,能实现极速生成。如果你正在构建智能体或多轮对话系统,并追求远超vLLM的推理性能,SGLang是理想选择。
