Créavores AEO Audit

What Créavores AEO Audit MCP server does

Créavores AEO Audit MCP server evaluates whether a web page can be discovered, understood, and cited by answer engines and AI agents. Its main audit divides scoring across crawlability, content structure, and E-E-A-T trust signals, then adds checks for llms.txt and access by AI crawlers. The full result includes a score from 0 to 100, a verdict, issues, and prioritized recommendations.

The server belongs in the 🔎 Search & Data Extraction category. It is useful for SEO and content reviews, technical audits, competitor benchmarking, and checking whether a page exposes enough structured information for automated systems to use it accurately.

How it works

The server fetches requested pages directly with an identifiable User-Agent and a 15-second timeout. It does not require an API key, configuration file, or environment variable. Each tool returns JSON, including an error object when a page or data source cannot be reached, rather than propagating an unhandled exception.

Créavores AEO Audit MCP server can inspect a URL for a 40–60 word answer directly below the H1, validate server-delivered JSON-LD, and assess page features relevant to agent navigation. It can also scan paragraphs, lists, tables, and FAQ answers to select the passage most likely to be quoted and explain the result.

Setup and configuration

Run the server with:

npx creavores-aeo-mcp

Claude Desktop can start it with the npx -y creavores-aeo-mcp command in its MCP server configuration. Claude Code can register the same command with claude mcp add or an .mcp.json file. No installation step beyond having the runtime needed for npx is described, and no environment variables are required.

Tools and capabilities

The available tools cover these tasks:

• Run a complete AEO audit with scores, findings, and recommendations.
• Check for a direct answer under the page H1.
• Extract JSON-LD types, parsing errors, common schema issues, and server-side delivery.
• Check an llms.txt file for its title, summary, section headings, and Markdown links.
• Review bylines, dates, citations, methodology links, legal notices, and Organization schema.
• Check agentic readiness through robots rules, headings, image alt text, ARIA landmarks, viewport metadata, and canonical URLs.
• Simulate AI citation selection and compare two or three URLs.
• Query French website-price cells and public-procurement aggregates from live datasets.
• Retrieve French editorial-calendar dates with sector, source, and content-angle fields.

Data responses include the source file URL, license, generation timestamp, and a citation page. The pricing and procurement tools support filters such as type, unit, and buyer type; the editorial tool supports date, sector, and result-limit filters.

Limitations and notes

The server analyzes fetched page content and reported web data; it does not claim to reproduce the behavior of every AI engine. A page or dataset that is unreachable produces an error JSON response. The editorial calendar material described in the source covers September 2026 through December 2027. Pricing and procurement results depend on the live referenced datasets and their available publishable cells.

The scoring logic is compatible with the public @macalc/aeo-score package and the associated online audit, according to the project documentation. Créavores AEO Audit MCP server is therefore best used as a diagnostic and comparison tool, not as a guarantee that an answer engine will cite a page.