AI Seo MCP

What AutomateLab-tech/ai-seo-mcp MCP server does

AutomateLab-tech/ai-seo-mcp MCP server evaluates whether public web pages are easy for AI systems to discover, interpret, extract from, and cite. Its composite page audit produces a score, grade, dimension-level results, and prioritized findings with suggested fixes. The checks cover schema.org markup, page structure, crawler permissions, entity signals, sitemap freshness, canonical and OpenGraph configuration, and related technical factors.

The server also supports focused checks and content operations. It can validate JSON-LD, inspect robots.txt and XML sitemaps, assess citation worthiness, identify named entities and sameAs links, generate or validate llms.txt files, and rewrite content for Answer Engine Optimization or Generative Engine Optimization. Page comparisons and citation simulations are available for more targeted analysis.

How it works

The server communicates with MCP clients through stdio. An MCP client calls one of the available tools with a URL or other supported input, and the server fetches and analyzes the relevant public content. audit_page combines multiple dimensions into one result, while audit_site combines page, robots, sitemap, and schema checks for a broader review.

By default, page audits read the returned HTML. For React, Vue, Angular, and other client-rendered applications, callers can request headless rendering with render: "headless". That mode uses Chromium and typically adds three to ten seconds per audit. Static rendering remains the default for pages whose useful content is present in the initial HTML.

Setup and configuration

AutomateLab-tech/ai-seo-mcp MCP server requires Node.js 20 or later. Run it with:

npx -y @automatelab/ai-seo-mcp

Claude Desktop can launch it using the npx command and the -y @automatelab/ai-seo-mcp arguments in its MCP server configuration. The same command-and-arguments pattern applies to MCP clients that support stdio, including Cursor and Windsurf. No API key or vendor credential is required.

Headless audits require the optional playwright-core package and a Chromium installation. The repository also provides a GitHub Action that audits configured URLs in a workflow, writes a job summary, and can fail when scores fall below a selected threshold or regress.

Tools and capabilities

The documented tool surface includes:

• Composite page, site, and sitemap audits with scores, grades, distributions, and prioritized findings.
• JSON-LD, canonical, technical HTML, robots.txt, XML sitemap, and llms.txt validation.
• Scoring for AI Overview eligibility, citation worthiness, agentic browsing, and test citation scenarios.
• Generation of llms.txt and optional llms-full.txt from a sitemap.
• AEO and GEO rewrites, entity extraction, page comparison, and Markdown report output.
• Reports saved under MCP_WORKSPACE_ROOT when using the report tool.

Limitations and notes

The audit targets public URLs and does not crawl private data. Raw-HTML mode can miss content that appears only after client-side JavaScript runs; use headless rendering for those pages. Citation and eligibility scores are assessments based on the checks and scoring methods described by the project, not guarantees that a search or AI system will cite a page. The optional GitHub Action and browser-rendering setup are separate from the basic stdio launch command.