Google Search Console MCP Server: Tools & Setup Guide

A Search Console MCP server lets an AI assistant read your Google Search Console data: clicks, impressions, rankings and indexing status. MCPifex hosts one for you. You connect the Google account that owns your site once, then any MCP client such as Claude, ChatGPT or Cursor can query it with one API key.

What you can ask it

Once it's connected, you skip the dashboard and ask in plain English. For example:

• "Which queries lost the most clicks this month compared with last month?" The AI calls compare_search_periods.
• "Which pages get lots of impressions but almost no clicks?" It uses get_search_by_page_query and sorts the result for you.
• "Is /pricing indexed, and if not, why?" It runs inspect_url_enhanced and reads Google's answer.

Every call is logged in the portal, so you can see exactly what was asked.

The tools

There are 19 tools as of September 2026. The read tools are on by default. Write and destructive tools are off until you switch them on for an instance.

Read (on by default)

• get_capabilities: what the server supports.
• list_properties: the Search Console properties your connected account can see.
• get_site_details: verification and permission details for one property.
• get_search_analytics: clicks, impressions, CTR and position for a date range.
• get_performance_overview: the overall performance trend.
• compare_search_periods: two date ranges side by side.
• get_search_by_page_query: performance by page and query.
• get_advanced_search_analytics: several dimensions and filters at once.
• inspect_url_enhanced: Google's URL Inspection for one URL.
• batch_url_inspection: URL Inspection for many URLs in one call.
• check_indexing_issues: common indexing problems on a property.
• get_sitemaps, list_sitemaps_enhanced and get_sitemap_details: your submitted sitemaps, with status and errors.

Write (off by default)

• submit_sitemap: submit a sitemap URL.
• manage_sitemaps: list, inspect, submit or delete sitemaps in one call.
• delete_sitemap: remove a submitted sitemap.

Destructive (off by default)

• add_site: add a property to the connected account.
• delete_site: remove a property from the connected account.

Safe by default

A fresh instance can only read. The gateway hides tools that are off from the AI's tool list and refuses any call to them, even if the AI asks by name. Turning on a write tool is a deliberate choice you make per instance in the portal.

What you'll need

There's no API key or JSON file to paste. You fill in two things in the portal when you create the instance:

• Google account: click Connect Google account and sign in. The account can be different from the one you use to log in to MCPifex. It must be an owner or full user of the Search Console property you want to query.
• Data state (optional): All matches the Search Console dashboard and is the default. Final only returns settled data, which runs two to three days behind.

When you click Test connection, the portal calls list_properties, so a missing permission shows up right away. To switch accounts later, connect again from the instance. The quickstart covers the rest of the setup.

Connect your AI app

First create a free account, set up the instance and generate an API key. It starts with mcpx_. Then replace <YOUR_MCPX_KEY> below with your key.

Claude Code

#!/bin/sh
# MCPifex — Claude Code CLI snippet.
# Replace <YOUR_MCPX_KEY> with your MCPifex API key (starts "mcpx_"), from
# the portal's API key page, then run this once from any shell that has the
# `claude` CLI on PATH.
claude mcp add mcpifex --transport http https://mcpifex.com/mcp \
  --header "Authorization: Bearer <YOUR_MCPX_KEY>"

Cursor

Add this to .cursor/mcp.json in your project, or to Cursor's global mcp.json, then reload Cursor:

{
  "mcpServers": {
    "mcpifex": {
      "type": "http",
      "url": "https://mcpifex.com/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_MCPX_KEY>"
      }
    }
  }
}

Claude Desktop

Use Claude's remote custom connector where it's available. Otherwise, this mcp-remote bridge works, and needs Node.js and npx. Merge the mcpifex entry into the mcpServers object in claude_desktop_config.json, then restart the app:

{
  "mcpServers": {
    "mcpifex": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://mcpifex.com/mcp",
        "--header",
        "Authorization:${MCPIFEX_AUTH_HEADER}"
      ],
      "env": {
        "MCPIFEX_AUTH_HEADER": "Bearer <YOUR_MCPX_KEY>"
      }
    }
  }
}

ChatGPT

ChatGPT can't send a header, so the key goes in the URL instead: https://mcpifex.com/mcp/<YOUR_MCPX_KEY>. The quickstart has the step-by-step ChatGPT setup. Treat that URL like a password.

Good to know

• If Google revokes access, for example because a Workspace admin blocks third-party apps or you remove MCPifex from your Google account, the instance shows a revoked connection. Connect the account again to fix it.
• Search Console itself lags by a few days, whatever Data state you pick.
• Each AI session gets its own isolated server process, and revoking a key stops new calls within about a minute.
• Tool calls are unlimited. Plans only limit how many instances you have. See pricing, or browse the marketplace for other hosted servers.