Google Searchconsole MCP

What lionkiii/google-searchconsole-mcp MCP server does

The lionkiii/google-searchconsole-mcp MCP server exposes Google Search Console data and operations as MCP tools for AI clients. It focuses on SEO analysis and site diagnostics rather than changing website content or Search Console settings. The server can list authenticated accounts and accessible properties, retrieve search performance metrics, inspect individual URLs, and list submitted sitemaps.

Its analytics features cover clicks, impressions, click-through rate, and average position. Queries can be examined by dimensions such as search query, page, country, device, or date. Additional tools identify high-impression queries with relatively low CTR, rank pages by selected performance metrics, compare two date ranges, separate branded from non-branded traffic, and show a keyword’s daily trend. Results can also be exported as CSV or JSON.

How it works

The server runs locally and communicates with Google Search Console through the Google API. Initial authentication uses a browser-based Google OAuth flow. After the user grants read-only Search Console access, OAuth tokens are stored under ~/.gsc-mcp/tokens/. Search Console data is queried directly from Google rather than routed through an intermediary.

The package includes OAuth credentials for the standard setup, so a Google Cloud project is not required. It also supports an alternative configuration using a user-managed OAuth client ID and secret, or a credentials file stored at ~/.gsc-mcp/credentials.json.

Setup and configuration

The lionkiii/google-searchconsole-mcp MCP server requires Node.js 18 or newer and a Google account with access to Search Console. Install the npm package globally, then run gsc-mcp-auth to open the login flow. Separate account aliases can be supplied for users who need to authenticate multiple Google accounts, and the authentication command can list configured aliases.

For Claude Desktop, configure an MCP server entry that runs npx with the package name and the -y option. The README also identifies Cursor, Windsurf, VS Code with Cline, and other MCP-compatible clients as supported. Client configuration details may differ, but the process is to point the client at the local MCP process.

Tools and capabilities

The lionkiii/google-searchconsole-mcp MCP server includes 13 tools:

• List Google accounts, Search Console properties, and submitted sitemaps.
• Query analytics by dimensions, search appearance, or search type.
• Inspect indexing state, crawl information, mobile usability, and rich-result details for a URL.
• Find high-impression, low-CTR query opportunities.
• Rank top pages by clicks, impressions, CTR, or position.
• Compare performance between two periods and identify gains or losses.
• Analyze branded versus non-branded queries.
• Track daily metrics for a chosen keyword.
• Export analytics results as CSV or JSON.

Limitations and notes

Access is read-only according to the authentication instructions, so the documented tools are for retrieving and analyzing Search Console information rather than submitting changes. The account used for OAuth must already have access to the relevant Search Console properties. The standard setup stores tokens locally, and the project states that it does not collect or transmit Search Console data to third parties.

Using custom OAuth credentials adds Google Cloud setup: a project, the Search Console API, and a desktop OAuth client are required. The available material does not document support for Search Console write operations, sitemap submission, URL indexing requests, or data sources outside Google Search Console.