gsc-mcp

Google Search Console MCP server with 32 tools covering search analytics, URL inspection, the Google Indexing API, Google Analytics 4, and cross-platform GSC+GA4 analysis. Built on Python 3.11+ and FastMCP.

The main workflow it enables: ask Claude "which pages on my site are crawled but not indexed?" then "submit them for indexing", end to end, no manual Google Search Console tabs.

Why this exists

Two solid open-source projects cover parts of this problem:

[AminForou/mcp-gsc](https://github.com/AminForou/mcp-gsc) (Python, 1k+ stars) has excellent search analytics, rich SEO tooling, and solid OAuth + Service Account auth. It does not include the Google Indexing API at all (no submit_url, no submit_batch). You can analyze your traffic but you cannot request reindexing.

[Suganthan-Mohanadasan/Suganthans-GSC-MCP](https://github.com/Suganthan-Mohanadasan/Suganthans-GSC-MCP) (Node.js/TypeScript) adds the Indexing API, but its submit_batch is a sequential for loop with a 100ms delay between requests, not a real HTTP batch. It also mixes plain-text and JSON outputs, hardcodes version strings that diverge from package.json, and has no retry logic.

This project takes the auth patterns and SEO tools from the first, the Indexing API scope from the second, and fixes the gaps in both.

Why Python instead of Go or Rust

The Google API Python client (google-api-python-client) is the official, best-maintained client for these APIs. It ships service.new_batch_http_request() natively, which is what makes true HTTP multipart batching possible without implementing the wire format by hand. The Go and Rust ecosystem for Google APIs relies on unofficial or auto-generated clients that do not expose this method. FastMCP also has first-class Python support with minimal boilerplate. The trade-off is startup latency (a few hundred ms vs near-zero for a compiled binary), but for a local MCP server called interactively by Claude that trade-off is irrelevant.

What changed vs the two inspirations

Tools (32)

Requirements

• Python 3.11+
• Google Cloud project with these APIs enabled:
• Google Search Console API
• Web Search Indexing API
• Google Analytics Data API (for GA4 tools)
• Credentials: OAuth Desktop app OR Service Account JSON
• For the Indexing API: your account or service account needs Owner-level access in Search Console (Full access is not enough)
• Indexing API default quota: 200 requests per day

Installation

uvx gsc-mcp

Or with pip:

pip install gsc-mcp

To run from source:

git clone https://github.com/FlorianBruniaux/google-search-console-mcp
cd google-search-console-mcp
python3 -m venv .venv && source .venv/bin/activate
pip install -e .

Configuration

Copy .env.example to .env and fill in the relevant variables.

OAuth (interactive, good for personal use)

export GSC_CREDENTIALS_PATH=/path/to/oauth-client-credentials.json
gsc-mcp

The first run opens a browser for Google login. The token is saved to your OS user data directory (~/.local/share/gsc-mcp/ on Linux, ~/Library/Application Support/gsc-mcp/ on macOS) as JSON files.

Service Account (recommended for automation and Claude Desktop)

export GSC_SERVICE_ACCOUNT_PATH=/path/to/service-account.json
export GSC_SKIP_OAUTH=true
gsc-mcp

GA4 setup

GA4 tools use the same Service Account as GSC. Two steps to enable them:

1. Open GA4 Admin, go to Property Access Management, and add the service account email (the client_email field in your SA JSON) with the Viewer role.
2. Set the GA4_PROPERTY_ID environment variable to your numeric property ID (e.g. 123456789, visible in GA4 Admin under Property Settings). The properties/ prefix is added automatically if omitted.

"env": {
  "GSC_SERVICE_ACCOUNT_PATH": "/path/to/service-account.json",
  "GA4_PROPERTY_ID": "123456789"
}

GSC-only users are not affected: the GA4_PROPERTY_ID check runs lazily on the first GA4 tool call, never at startup.

To query a different property without changing the config, pass property_id directly to any GA4 or cross tool:

ga4_traffic_sources(property_id="443684366")
traffic_health_check(site="sc-domain:example.com", property_id="443684366")

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "gsc-mcp": {
      "command": "python",
      "args": ["-m", "gsc_mcp.server"],
      "env": {
        "GSC_SERVICE_ACCOUNT_PATH": "/absolute/path/to/service-account.json",
        "GSC_SKIP_OAUTH": "true"
      }
    }
  }
}

Or, once published to PyPI:

{
  "mcpServers": {
    "gsc-mcp": {
      "command": "uvx",
      "args": ["gsc-mcp"],
      "env": {
        "GSC_SERVICE_ACCOUNT_PATH": "/absolute/path/to/service-account.json",
        "GSC_SKIP_OAUTH": "true"
      }
    }
  }
}

Development

python3 -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"
pytest tests/ -v

167+ tests, all mocked (no real Google API calls needed).

License

MIT