SEO MCP Documentation

Learn how to connect an AI agent to real SEO crawl data, run evidence-based audits, and build safe technical SEO workflows with SEO MCP.

SEO MCP is a Model Context Protocol server for SEO work. It gives compatible AI agents structured access to your projects, crawls, pages, links, images, uptime monitoring, and Core Web Vitals.

Instead of pasting exports into a chat, you can ask a focused question. The agent discovers the correct project and crawl, retrieves only the relevant records, and explains its findings with the underlying evidence.

What SEO MCP is—and is not

SEO MCP is the connection between an AI agent and real SEO data. It does not replace a crawler, guarantee rankings, or give an agent unrestricted access to your account.

Start here

Connect your AI client

Authorize a compatible MCP client with OAuth and PKCE.

Run your first SEO audit

Move from a plain-language question to verified crawl findings.

Explore SEO workflows

Use task-focused playbooks for links, images, CWV, and crawl changes.

Browse all tools

Understand the 23 project, crawl, on-page, and monitoring tools.

What can SEO MCP analyze?

Area	Questions you can answer	Main tools
Projects	What sites can I access? Which project needs attention?	`list_projects`, `get_project_overview`, `compare_projects`
Crawls	What changed between releases? Is a crawl complete?	`list_crawls`, `get_crawl_summary`, `compare_crawls`
Pages	Which pages have 4xx responses, missing titles, or canonical problems?	`list_pages`, `get_page_details`
Links	Where do broken links appear? Which pages link to a target?	`list_links`, `get_link_sources`
Images	Which images are large, broken, or missing alt text?	`list_images`, `get_image_sources`
Experience	Which pages have weak Core Web Vitals or uptime signals?	`list_core_web_vitals`, `get_uptime_summary`

The safe default workflow

Call list_projects and select a valid project UID.
Call list_crawls and select a valid crawl UID for crawl-level work.
Use a summary tool to understand scope.
Use filtered record tools to verify the finding.
Explain impact, evidence, and the next action separately.
Change state only when the user explicitly requests it.

This order matters. It prevents guessed identifiers, reduces unnecessary data retrieval, and makes the final answer easier to audit.

How these docs are written

Every workflow begins with the direct answer, then expands into requirements, tool sequence, prompt examples, validation, and limitations. The content reflects the current SEO MCP server surface and is maintained by the Screpy product team.

Last reviewed: July 29, 2026.

Next step

Go to Connect your AI client to add the SEO MCP endpoint, or open the prompt library if you are already connected.