Free · 13 tools · MIT

AI-SEO MCP

A free Model Context Protocol server that gives Claude, Cursor, and any MCP-compatible agent 13 tools to audit, score, and rewrite pages for AI-citation eligibility. Built for content teams who want their pages to show up inside ChatGPT, Perplexity, Google AI Overviews, and Claude with web access, not just on page one of Google.

npx -y @automatelab/ai-seo-mcp

Get it on GitHub npm package Read the launch post

What it gives your agent

Thirteen tools, grouped into five families. Every tool returns the same structured finding shape: severity, category, where on the page, what to change, estimated impact. Auditors get scores plus a prioritized fix list. Rewriters use MCP sampling on the host model, with a graceful fallback to prompt templates when sampling is not available.

Tool	Family	Purpose
`audit_page`	audit	Composite AI-SEO audit. Eight dimensions (schema, technical, structure, robots, freshness, authority, entity density, sitemap) into a 0-100 score plus a ranked fix list.
`audit_schema`	audit	Validate JSON-LD against Schema.org rules and AI-citation best practice. Flags deprecated patterns.
`audit_canonical`	audit	Canonical link integrity, trailing-slash hygiene, og:url consistency.
`check_robots`	check	Parse robots.txt and report per-crawler allow/disallow for 10+ AI crawlers. Surfaces the GPTBot-blocked-but-OAI-SearchBot-allowed trap.
`check_sitemap`	check	Validate XML sitemaps: presence, URL count, lastmod freshness, image/video extensions.
`check_technical`	check	HEAD tag audit: canonical, OpenGraph, Twitter Card, hreflang, HTTPS, noindex, title hygiene.
`score_ai_overview_eligibility`	score	Score a page's probability of appearing in Google AI Overviews using current correlation factors.
`score_citation_worthiness`	score	Score how citable a page or text block is for Perplexity, ChatGPT, Google AI Overviews, and Claude.
`generate_llms_txt`	llms.txt	Generate llms.txt (and optionally llms-full.txt) from a domain's sitemap.
`validate_llms_txt`	llms.txt	Lint an existing llms.txt for spec compliance and broken links.
`extract_entities`	entities	Extract named entities, sameAs links, and citation-density score from a page.
`rewrite_for_aeo`	rewrite	Rewrite content for Answer Engine Optimization. BLUF structure, FAQ format, schema additions.
`rewrite_for_geo`	rewrite	Rewrite content for Generative Engine Optimization. Entity definitions, comparison tables, synthesis-ready structure.

What makes it different

Built for AI search, not classic SEO

Lighthouse will not flag missing FAQPage schema. Search Console will not tell you GPTBot is allowed but OAI-SearchBot is blocked. Ahrefs will not score citation worthiness. This MCP audits the signals AI assistants actually use to decide who to cite.

Deterministic rubrics, not opaque scores

Every score is the sum of explicit, published checks. Every finding carries a severity, a category, the exact location on the page, the fix to apply, and an impact estimate. If you do not agree with a finding, the rule is visible and editable.

Vendor-agnostic, no API keys

Audits ChatGPT, Perplexity, Google AI Overviews, Claude, and Microsoft Copilot citation signals from the same toolset. No registration. No paid-tier gates. Works the day you install it.

Polite by default

Every request goes through one fetch path that respects robots.txt, identifies itself honestly, sleeps between requests to the same host, and caps response size. The MCP is an auditor, not a scraper.

10+ AI crawlers covered

GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, anthropic-ai, PerplexityBot, Perplexity-User, Google-Extended, Applebot-Extended, Bytespider, Meta-ExternalAgent. Updated as Anthropic, OpenAI, Google, and Perplexity publish new agents.

Rewriters use the host model

rewrite_for_aeo and rewrite_for_geo use MCP sampling, so the rewrite is done by whichever model your client speaks to (Claude, GPT, Llama). The MCP supplies the rubric and the constraints; your model does the writing.

Install in 3 steps

Install with npx (Node 20 or later). npx -y @automatelab/ai-seo-mcp
Add the server to your MCP host config (~/.cursor/mcp.json for Cursor, claude_desktop_config.json for Claude Desktop).
```
{
  "mcpServers": {
    "ai-seo": {
      "command": "npx",
      "args": ["-y", "@automatelab/ai-seo-mcp"]
    }
  }
}
```
No API keys required. All five env vars (USER_AGENT, FETCH_TIMEOUT_MS, MAX_BYTES, RESPECT_ROBOTS, INTER_REQUEST_DELAY_MS) are optional with sensible defaults.
Restart your MCP host. The 13 tools appear in the MCP panel, ready to use.

Works with Claude Desktop, Claude Code, Cursor, Cline, Continue, and any MCP-compatible agent harness. Full reference: the GitHub readme.

Example workflow

Ask Claude: "Run an AI-SEO audit on https://example.com/my-post and tell me the top three things to fix." Claude calls audit_page, gets back an eight-dimension score with prioritized findings, and reports them in order of impact. For a "missing FAQPage schema" finding, Claude can then call rewrite_for_aeo on a passage and return a citation-ready answer block with the JSON-LD wrapper already applied. The whole loop runs without any API keys, against any public URL.

FAQ

What is the AI-SEO MCP?

A free Model Context Protocol server that gives Claude, Cursor, and any MCP-compatible agent 13 tools to audit, score, and rewrite pages for AI citation eligibility. It looks at the signals AI assistants use to decide what to cite: schema completeness, FAQ structure, AI-crawler allowlists in robots.txt, llms.txt, entity density, freshness, and authority.

Do I need an API key?

No. The MCP fetches public URLs directly. No registration, no paid tiers, no rate limits beyond the polite-fetch defaults (one request per 1.5 seconds per host, 15-second timeout, 5 MB response cap). Override the defaults via env vars if you need to.

How is this different from Lighthouse, Search Console, or Ahrefs?

Those tools measure classic search ranking factors. The AI-SEO MCP measures the signals that decide whether AI assistants cite your page: FAQPage schema, AI-crawler allowlists in robots.txt (GPTBot vs OAI-SearchBot can have different rules on the same domain), llms.txt presence and validity, sameAs entity links, and the structure generative models extract answers from.

Does it actually rewrite content, or just suggest changes?

Both. Auditors return findings and fix recommendations. Rewriters (rewrite_for_aeo and rewrite_for_geo) use MCP sampling so the host model - Claude, GPT, or whichever your client uses - performs the rewrite under the rubric the MCP supplies. If your client does not support sampling yet, the MCP falls back to returning a prompt-template output the agent can run inline.

Which agents does it work with?

Anything that speaks MCP: Claude Desktop, Claude Code, Cursor, Cline, Continue, VS Code with Copilot, and any MCP-capable harness. Configuration is the same JSON block in each host's MCP config file.

What does it cost?

The server is MIT-licensed and free. You pay only for your AI agent's model usage. There are no per-audit fees, no rate-limited free tier, no upgrade prompt.

Want it wired into your publishing pipeline?

We use the AI-SEO MCP to audit every post we ship. If you want it set up against your CMS, sitemap, or content workflow, or a full AI-SEO audit done end to end on a site you own, we can do that.

Get in touch