Your Website Intelligence, Available Everywhere

How It Works

The Model Context Protocol (MCP) is an open standard that lets AI tools access external data sources through a consistent interface. Evergreen ships a fully featured MCP server that exposes your project data — crawl results, content audit scores, site structure, and correlated analytics — to any compatible client.

When you connect an MCP client to your Evergreen server, your AI assistant gains the ability to read your website intelligence data, answer questions about your site, and incorporate real content health metrics into its responses. The data stays in Evergreen; the AI tool queries it on demand.

Setup Steps

1. Generate an API key in your Evergreen account under Settings > API Keys.
2. Copy your MCP server URL from the Integrations tab — it follows the format https://mcp.evergreen.com/v1/{project-id}.
3. Add the server to your MCP client configuration. For Claude Code, add it to your .claude/settings.json. For Cursor, add it to the MCP section in your project settings.
4. Authenticate with your API key when prompted by the client.
5. Start querying. Your AI tool now has access to your Evergreen project data.

Configuration is a one-time setup per project. Once connected, the MCP server stays available across sessions.

What Data You Get

The Evergreen MCP server exposes a comprehensive set of tools and resources that AI clients can use:

• Site Structure — Query the full crawled sitemap, including URL hierarchy, depth, internal link relationships, and navigation paths.
• Content Health Scores — Access per-page audit results covering titles, meta descriptions, headings, image alt text, word count, readability, and overall health grades.
• Crawl Issues — Retrieve lists of broken links, redirect chains, missing metadata, duplicate content, and other technical problems the crawler detected.
• Analytics Data — When GA4 and Search Console are connected, query page-level traffic, engagement metrics, keyword rankings, and CTR data through the same MCP interface.
• Project Summaries — Get high-level overviews of site health, page counts, issue breakdowns, and trend data across crawls.

All data is returned in structured formats that AI tools can reason over, filter, and incorporate into larger workflows.

Why It Matters

Website intelligence is most valuable when it meets you where you work. Dashboards are useful, but they require you to context-switch. The MCP integration eliminates that friction by bringing Evergreen data directly into the tools you already use.

For Content Strategists

Ask your AI assistant to "list all pages with a content health score below 50 that received more than 500 sessions last month." Get a prioritized list without opening a single dashboard. Draft content briefs grounded in real audit data. Generate reports that cite actual metrics instead of placeholders.

For Developers

Query Evergreen from your editor to check whether a page you are refactoring has SEO dependencies. Ask which internal links point to a URL you are about to deprecate. Surface crawl errors related to the section of the site you are working on — all without leaving your coding environment.

For SEO Teams

Build automated workflows that combine Evergreen data with other MCP-connected tools. For example, query Evergreen for pages with high impressions but low CTR, then pass those results to an AI writing assistant to draft improved title tags and meta descriptions. The MCP protocol makes Evergreen composable with the rest of your toolchain.

An Open Protocol for the Future

MCP is not proprietary to any single vendor. As the ecosystem grows, your Evergreen data becomes accessible to an expanding set of tools — from IDE extensions and CLI agents to custom scripts and automation platforms. Connecting once means your website intelligence is ready for whatever comes next.

Your website data should not be locked behind a single interface. Set up the Evergreen MCP server and put your website intelligence to work wherever your team already operates.