← 返回 Skills 市场
821
总下载
2
收藏
2
当前安装
1
版本数
在 OpenClaw 中安装
/install geo
功能描述
Optimize for AI recommendations so ChatGPT, Claude, and Perplexity suggest your brand first.
使用说明 (SKILL.md)
Triggers
Activate on: AI optimization, generative engine optimization, AI recommendations, ChatGPT ranking, get recommended by AI, AI visibility, AI search optimization.
What Is GEO
Generative Engine Optimization = making AIs recommend you when users ask for solutions in your category.
Unlike SEO (ranking in search results), GEO is about being in the AI's training data and knowledge in a way that makes you the natural answer.
The Audit Process
Before optimizing, measure where you stand:
- Run simulation queries across multiple AI models
- Test different phrasings: "best X", "recommend X", "which X should I use", "X for [use case]"
- Test different contexts: beginner vs expert, specific needs, budget constraints
- Document which brands get mentioned and in what position
See audit.md for a complete audit workflow with example prompts.
Why AIs Recommend What They Recommend
| Factor | Weight | How to Influence |
|---|---|---|
| Training data frequency | High | Get mentioned everywhere (articles, Reddit, forums) |
| Authority signals | High | Expert endorsements, official docs, Wikipedia |
| Recency | Medium | Fresh content in sources AI scrapes |
| Specificity | Medium | Be THE answer for a niche, not A answer for everything |
| User context | Variable | Match your positioning to query patterns |
Core Strategies
1. Own Your Category Description
When someone asks "what is [your category]?" — your brand should be in the example or definition.
- Get mentioned in Wikipedia articles for your category
- Author definitive guides that AIs cite
- Be in "best of" lists that get scraped
2. Dominate Discussion Platforms
Reddit, HN, Stack Overflow, Quora = AI training goldmines.
- Authentic presence (not spam)
- Be the helpful answer, not the promoted one
- Build reputation accounts that mention your product naturally
3. Technical Content Authority
AIs weight documentation and technical content heavily.
- Comprehensive public docs
- Integration guides with popular tools
- Open source presence (GitHub stars, contributors)
4. Strategic PR and Mentions
- Get in TechCrunch, Product Hunt, industry publications
- Expert interviews and podcasts (transcripts get indexed)
- Case studies with named customers
See strategies.md for detailed playbooks by business type.
Testing Your Progress
Run the same audit queries monthly. Track:
- Position (mentioned 1st, 2nd, 3rd, or not at all)
- Context (when are you recommended vs competitors)
- Phrasing sensitivity (which queries favor you)
Red Flags
- ❌ AI never mentions you → no presence in training data
- ❌ AI mentions you negatively → reputation problem in sources
- ❌ AI confuses you with competitor → brand differentiation issue
- ❌ AI only mentions you for wrong use case → positioning mismatch
What This Doesn't Cover
- Traditional SEO (Google rankings) → see
seoskill - Paid advertising → see
adsskill - Social media marketing → see relevant platform skills
安全使用建议
This is a coherent marketing/playbook skill (not code) explaining how to increase the chance AIs recommend your brand. It doesn't request secrets or install anything. Two practical cautions before you act: (1) If you want the agent to automate multi-model audits, you'll need to provide API credentials for each service — avoid granting account keys or posting on your behalf unless you trust the agent and have reviewed limits and policies. (2) Some recommended tactics (Wikipedia edits, forum activity, mass posting) can violate site rules or be seen as astroturfing; follow platform policies and prioritize authentic, high-quality contributions to avoid reputational or policy risk.
功能分析
Type: OpenClaw Skill
Name: geo
Version: 1.0.0
The skill bundle describes a Generative Engine Optimization (GEO) process. While the overall intent appears benign, the `audit.md` file contains explicit 'Automation Tips' instructing the OpenClaw agent to 'Run the following queries on [model]' across multiple external AI services (ChatGPT, Claude, Perplexity, Gemini, Copilot) and 'Output as JSON'. This grants the agent direct network communication capabilities to external endpoints. Although this is plausibly needed for the stated purpose of auditing AI recommendations, it represents a high-risk capability that could be exploited via prompt injection or other means to perform unintended external communication or resource consumption, classifying it as suspicious due to the inherent vulnerability risk rather than clear malicious intent.
能力评估
Purpose & Capability
The name/description (GEO: generative engine optimization) match the SKILL.md, audit.md, and strategies.md content. All recommended actions (audits, content, PR, forum participation) are coherent with the stated goal and no unrelated credentials, binaries, or system paths are requested.
Instruction Scope
Instructions stay within marketing/AIOps scope: designing queries, testing multiple models, documenting results, and producing content. The docs mention automating audits (sub-agents, running queries across models) which would in practice require access to external AI services/accounts, but the skill does not itself request credentials or direct reading of unrelated local files. This is a scope note (automation implies additional credentials/privileges if you choose to implement it).
Install Mechanism
No install spec and no code files — instruction-only skill. Nothing is written to disk or downloaded, which is the lowest-risk install surface.
Credentials
The skill declares no environment variables, secrets, or config paths and its guidance doesn't require access to system credentials to be useful. Be aware that if you follow the automation tips (running queries against paid AI models), you'll need API keys/accounts — those are not requested by the skill itself.
Persistence & Privilege
always is false and the skill does not request persistent system presence or attempt to modify other skills or agent-wide settings. Autonomous invocation is allowed by default but the content does not require permanent privileges.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install geo - 安装完成后，直接呼叫该 Skill 的名称或使用
/geo触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
Initial release
元数据
常见问题
GEO 是什么？
Optimize for AI recommendations so ChatGPT, Claude, and Perplexity suggest your brand first. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 821 次。
如何安装 GEO？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install geo」即可一键安装，无需额外配置。
GEO 是免费的吗？
是的，GEO 完全免费（开源免费），可自由下载、安装和使用。
GEO 支持哪些平台？
GEO 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 GEO？
由 Iván（@ivangdavila）开发并维护，当前版本 v1.0.0。
推荐 Skills