Google E Commerce (GA4, Search Console, Merchant Center)

What Google E-commerce (GA4, Search Console, Merchant Center) MCP server does

Google E-commerce (GA4, Search Console, Merchant Center) MCP server gives an AI assistant read access to the Google services commonly used to operate an online shop. It covers Google Analytics 4, Search Console, Merchant Center, Tag Manager, the Web Search Indexing API, and PageSpeed Insights.

The server is designed for investigation rather than administration. It does not create, update, publish, or delete resources through its tools. Typical questions include how much traffic came from organic search, which products have Merchant Center warnings, whether a URL is indexed, whether a page has been crawled, and whether analytics tracking appears more than once.

How it works

The server uses a Google OAuth desktop application for authorization. After consent, it sends requests from the local machine directly to Google. The returned data is formatted for model use, including report totals, metric units and definitions, and dates interpreted using the relevant GA4 property timezone. Reports also indicate when recent data may still change.

A status tool reports which services are configured and whether the token works. GA4 tools can enumerate accessible accounts and properties, query reports, and inspect the last 30 minutes of activity. Search Console tools cover search performance, URL inspection, and sitemaps. Merchant tools retrieve data sources, product issues, and Merchant Query Language reports. Additional tools inspect Tag Manager containers, submitted URL indexing status, and PageSpeed results.

Setup and configuration

Create a Google Cloud project, enable the APIs required for the services you plan to use, configure an OAuth consent screen, and create a Desktop app OAuth client. The documented APIs include the Google Analytics Data and Admin APIs, Search Console API, Merchant API, Tag Manager API, Web Search Indexing API, and PageSpeed Insights API. Merchant API use also requires registering the Cloud project with the Merchant Center account.

The package can be installed with uvx or pip. Its setup command accepts the OAuth client secret, opens the Google consent flow, and stores the resulting token in the operating system keyring. The project documents Windows Credential Manager, macOS Keychain, and Secret Service as supported keyring locations. A check command is available to verify the setup.

Client configuration can provide identifiers such as GA4_PROPERTY_ID, GSC_SITE_URL, MERCHANT_ACCOUNT_ID, and GTM_CONTAINER_ID. PAGESPEED_API_KEY can be supplied for PageSpeed Insights; the documentation describes this as optional because the anonymous quota may be exhausted. Claude Desktop and Claude Code configuration examples are included in the repository.

Tools and capabilities

• server_status: check service configuration and token validity.
• ga4_report: query GA4 reports such as channels, landing pages, purchases, revenue, and periods.
• ga4_realtime and ga4_properties: inspect recent activity and find readable properties.
• gsc_performance, gsc_inspect_url, and gsc_sitemaps: review search metrics, indexing details, canonical selection, crawl information, and sitemap status.
• merchant_data_sources, merchant_product_issues, and merchant_report_query: inspect feeds, countries, fetch URLs, disapprovals, warnings, and Merchant Query Language results.
• gtm_inventory, indexing_status, and pagespeed: review Tag Manager resources, submitted URL knowledge, Lighthouse scores, SEO scores, and Core Web Vitals.

Limitations and notes

Google E-commerce (GA4, Search Console, Merchant Center) MCP server is read-only, so it cannot fix product disapprovals, submit changes, publish tags, or alter Google properties. Access depends on the Google account permissions granted during OAuth and on enabling the corresponding APIs.

Some services need additional setup. Merchant Center requires project registration, while PageSpeed may benefit from an API key because its shared anonymous quota can be depleted. The server supports Python 3.10 or newer according to the repository metadata. Review the pinned installer scripts before running them if using the supplied installation path.