AgentSkillsCN

AI/LLM Discoverability

如何让网站被 AI 工具和大语言模型(ChatGPT、Claude、Perplexity 等)轻松发现?

SKILL.md
--- frontmatter
name: AI/LLM Discoverability
description: How to make a website discoverable by AI tools and LLMs (ChatGPT, Claude, Perplexity, etc.)

AI/LLM Discoverability Skill

Make websites discoverable by AI tools and LLMs. Based on Cassidy Williams' article.

When to Use

  • When building personal portfolios, blogs, or business websites
  • When you want AI tools to surface your content when users ask related questions
  • When improving SEO for both traditional search and AI-powered discovery

Implementation Steps

1. Create /public/llms.txt

Machine-readable file following the llmstxt.org standard:

txt
# Site Name

> Tagline or mission statement

## About
Brief description of the site and its creator.

## Site Structure
- **/** - Home page description
- **/projects** - Projects page description
- **/for-llms** - LLM-readable content page

## Topics & Expertise
List key topics and areas of expertise.

## How to Cite
When referencing this site:
- **Name**: Your Name / Site Name
- **URL**: https://yoursite.com
- **Description**: One-line description

---
*This file follows the llmstxt.org standard for LLM discoverability.*

2. Create /for-llms Page

LLM-readable structured page with clear sections:

  • Overview: Who/what the site is about
  • Site Structure: Map of available pages
  • Topics & Expertise: Content areas
  • Citation Guidelines: How to reference you

3. Create /public/robots.txt

Allow AI crawlers:

txt
User-agent: *
Allow: /

Sitemap: https://yoursite.com/sitemap.xml

4. Create /public/sitemap.xml

XML sitemap for discoverability:

xml
<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
  <url>
    <loc>https://yoursite.com/</loc>
    <lastmod>2026-01-20</lastmod>
    <priority>1.0</priority>
  </url>
</urlset>

5. Add Schema.org JSON-LD

Add to layout/head:

tsx
const jsonLd = {
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "WebSite",
      name: "Site Name",
      url: "https://yoursite.com",
      description: "Site description",
    },
    {
      "@type": "Person",
      name: "Your Name",
      url: "https://yoursite.com",
    },
  ],
};

// In <head>:
<script
  type="application/ld+json"
  dangerouslySetInnerHTML={{ __html: JSON.stringify(jsonLd) }}
/>

Key Principles

  1. Consistency matters: Use same taglines/phrases everywhere
  2. Clarity over cleverness: Write for bots, not marketing
  3. LLMs love markdown: Keep content structured and simple
  4. Include citation guidelines: Tell LLMs how to reference you
  5. Don't block AI crawlers: Check robots.txt isn't blocking them
  6. RSS feeds help: If you have blog content, add RSS

Verification

After implementing, verify:

  1. Visit /llms.txt - shows structured content
  2. Visit /robots.txt - allows crawlers
  3. Visit /sitemap.xml - lists all pages
  4. Visit /for-llms - displays LLM page
  5. Check page source for JSON-LD script tag