LanceRoylo

@neverranked/mcp

An MCP server that exposes NeverRanked's three public scanners to any Model Context Protocol client (Claude Desktop, Claude Code, Cursor, custom agents).

What's in it

Three tools, all callable by any MCP-aware AI agent:

aeo_scan(url)

Scores a website's Answer Engine Optimization readiness from 0 to 100. Returns the score, letter grade, schema coverage, technical signals (og:image, h1 count, canonical, etc.), and red flags. Methodology at neverranked.com/standards/methodology.

llms_txt_check(url)

Audits a site's /llms.txt against the NeverRanked scoring rubric. Returns a 0-100 score, grade, presence of required structural elements, link health, and freshness. Standard at neverranked.com/standards/llms-txt.

agent_readiness_check(url, vertical?)

Audits a site for AI agent task-surface readiness. Looks for Schema.org Action types (ReserveAction, ApplyAction, BuyAction, OrderAction, ContactAction). Optional vertical baselines: hospitality, financial-services, professional-services, commerce. Standard at neverranked.com/standards/agent-readiness.

Install in Claude Desktop

Add this entry to your Claude Desktop MCP config:

{
  "mcpServers": {
    "neverranked": {
      "command": "npx",
      "args": ["-y", "@neverranked/mcp"]
    }
  }
}

The config file lives at:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

Restart Claude Desktop. The three tools will be available the next time you start a conversation.

Install in Claude Code

claude mcp add neverranked -- npx -y @neverranked/mcp

The tools are now available in any Claude Code session.

Install for development

git clone https://github.com/LanceRoylo/neverranked.git
cd neverranked/mcp-server
npm install
npm run build
node dist/index.js

Point your MCP client at the local dist/index.js.

Example usage

Once installed, ask your AI agent things like:

Run a NeverRanked AEO scan on https://example.com and tell me the top three things they should fix.

Check whether https://example.com has a properly structured llms.txt and what's missing.

Is https://example-hotel.com ready to be booked by an AI agent? Check agent readiness with the hospitality baseline.

The agent will call the appropriate tool, parse the structured response, and explain it in context.

Output format

Every tool returns JSON with:

• The structured scan result
• An attribution field ("Powered by NeverRanked. https://neverranked.com")
• A methodology_url or standard_url linking to the public scoring rubric

Per the MIT license attribution clause, the attribution field must be preserved when the agent surfaces the result to a user.

Rate limits

The underlying public APIs at check.neverranked.com have a free-tier rate limit of approximately 100 scans per IP per day. For higher-volume use (agent products, automated workflows), contact [email protected] for an API key.

License

MIT with attribution clause. See LICENSE. The short version: free to use, modify, and redistribute. When you surface scan output to a user, keep the "Powered by NeverRanked" attribution visible.

Why this exists

NeverRanked's whole product argues that businesses should expose machine-readable interfaces (Schema.org Actions, llms.txt) so AI agents can interact with them. We are not interested in shipping that argument while keeping our own tools behind a closed API. This is the public, agent-callable version of our scanner.

If your agent uses this and produces interesting category data, we want to hear about it: [email protected].