MCP Server

aeo.js ships an MCP server so AI agents — Claude Code, Claude Desktop, Cursor, or anything MCP-compatible — can audit sites and generate AEO files conversationally.

Setup

No install needed. Add to your MCP client config:

{
  "mcpServers": {
    "aeo": {
      "command": "npx",
      "args": ["-y", "aeo.js", "mcp"]
    }
  }
}

For Claude Code:

claude mcp add aeo -- npx -y aeo.js mcp

Tools

audit_url

Audit any live site for AI answer-engine readiness. Crawls the AEO surface (robots.txt, llms.txt, sitemap.xml, ai-index.json) plus the homepage and up to 10 inner pages, and returns the 0–100 GEO readiness score, the AI crawler access matrix, citability scores, and top fixes.

"Audit example.com and tell me why ChatGPT can't see it."

score_citability

Score draft content (markdown or plain text) for AI citability — how likely answer engines are to extract and cite it. Returns a 0–100 score across four dimensions with concrete hints. Useful while writing: ask your agent to score and revise a page until it's citable.

"Score this blog post for AI citability and rewrite the weakest sections."

generate_aeo_files

Generate the full AEO file set (robots.txt, llms.txt, llms-full.txt, sitemap.xml, ai-index.json, schema.json, per-page markdown) into the current project's output directory.

"Generate AEO files for this project, title 'My Site', url https://mysite.com."

Notes

• The server speaks newline-delimited JSON-RPC over stdio; logs go to stderr.
• audit_url requires network access; the other tools work offline.
• Requires Node 18+.