AI Readiness

What epistemedeus/ai-readiness does

The epistemedeus/ai-readiness MCP server audits a website’s homepage for technical signals that affect discoverability and interpretation by AI search systems. It returns a score from 0 to 100, a letter grade, and a specific suggested fix for each failed or incomplete check.

The audit covers whether common AI and search crawlers are allowed by robots.txt, including GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended, CCBot, and Bingbot. It also looks for JSON-LD structured data, title and meta description quality, Open Graph tags, an XML sitemap, and llms.txt. The README describes llms.txt as a minor hygiene check with no proven effect on AI citations.

How it works

The package runs locally as a dependency-free Node.js program and communicates with MCP clients over stdio. It does not require a signup, API key, or hosted service. An MCP client can call check_ai_readiness with a URL to receive the audit result inline.

The generate_ai_readiness_fixes tool uses information from the site to produce starter Organization and FAQPage JSON-LD, along with an AI-crawler-friendly robots.txt. These outputs are starting points for implementation rather than a deployment mechanism for changing the website.

The same project also provides a command-line mode for direct checks and a --json option for scripting. A GitHub Action is documented for checking a URL during deployments, but that workflow is separate from the local MCP transport.

Setup and configuration

Run the MCP server with Node.js 18 or newer:

npx -y github:epistemedeus/ai-readiness mcp

A compatible client can use the following stdio configuration:

{
  "mcpServers": {
    "ai-readiness": {
      "command": "npx",
      "args": ["-y", "github:epistemedeus/ai-readiness", "mcp"]
    }
  }
}

The README specifically mentions Claude Desktop, Cursor, and Cline as compatible MCP clients. The epistemedeus/ai-readiness MCP server has no required environment variables or credentials.

Tools and capabilities

• check_ai_readiness(url): checks the supplied website and returns its score, grade, findings, and fixes.
• generate_ai_readiness_fixes(url): creates starter Organization and FAQPage JSON-LD plus a robots.txt intended to allow the listed AI crawlers.
• Command-line JSON output: supports --json for automation outside MCP.
• Deployment checks: a documented GitHub Action can check a URL on deploy.

Limitations and notes

The epistemedeus/ai-readiness MCP server checks markup and crawler-access fundamentals; it does not determine whether ChatGPT, Perplexity, or Google AI will actually cite a site for specific buyer queries. It also does not provide competitor comparisons or live citation testing.

The package includes http.js, but the README identifies it as a minimal stateless JSON-RPC-over-HTTP POST adapter rather than MCP Streamable HTTP. It is not described as a production remote MCP transport, and there is no hosted remote MCP URL. The local stdio configuration is therefore the supported MCP setup described for this package.

The checker reports llms.txt when present, but the project explicitly cautions that this file has no proven citation effect. Generated fixes must still be reviewed and added to the target website by its owner.

Getting started with this epistemedeus/ai-readiness MCP server

Always refer to the official documentation for the most accurate and up-to-date information.