The searchconsole MCP server provides read-only access to your Google Search Console data via natural language prompts. It exposes tools that let you list all verified properties, query analytics data (like clicks, impressions, CTR, and average position) with flexible filters, and retrieve sitemaps for a given property. You can connect to the MCP from compatible clients (including Claude and other MCP clients) over stdio, allowing your AI assistant to answer questions about your site's search performance without exposing your credentials or raw data to the client.

After starting the server, you can ask questions such as: which keywords drove the most clicks this month, which pages have high impressions but low CTR, what are my sitemaps and their statuses, or what properties I own and what access levels I have for each. The tools map directly to common Search Console capabilities:

• list_sites: enumerate verified properties and permission levels
• query_search_analytics: fetch keyword/page performance with filters for date ranges, dimensions, and pagination
• list_sitemaps: retrieve submitted sitemaps for a property These tools enable natural-language queries while the underlying data access remains read-only, helping you monitor performance safely.

Prerequisites:

• Python 3.11 or newer installed on your system
• Internet access to install dependencies
• (Optional) uv (uvx) installed if you plan to run via uvx; instructions assume you will install via a standard Python package setup

Step-by-step:

1. Clone the repository: git clone https://github.com/chrishart0/searchconsole-mcp.git cd searchconsole-mcp

2. Create and activate a Python virtual environment (recommended): python3 -m venv venv source venv/bin/activate # on Unix or MacOS .\venv\Scripts\activate # on Windows

3. Install the MCP package dependencies (if a setup script or requirements is provided): pip install -r requirements.txt # if a requirements file exists

   or install the package in editable mode if applicable:

   pip install -e .

4. Run the MCP server using uvx (Python/uv) as shown in the Quick Start: uvx run --directory /path/to/searchconsole-mcp searchconsole-mcp

5. If you prefer npx/Node-based workflows or different deployment options, follow any alternative instructions in the repository README, but for this server the Python/uvx route is the intended path.

6. (Optional) Set Google credentials when required by your environment: export GOOGLE_APPLICATION_CREDENTIALS=/path/to/key.json or use gcloud auth application-default login for user credentials.

Test by connecting a client (e.g., Claude) and sending a simple prompt like: "List my Search Console properties" to verify connectivity.

Notes and tips:

• This MCP server uses Google Application Default Credentials. Ensure the service account or user has access to the Search Console properties you intend to query.
• When connecting from Claude or another client, you may need to provide environment variables (e.g., GOOGLE_APPLICATION_CREDENTIALS) or authentication steps per client documentation.
• If you encounter permission errors, double-check that the credentials have at least read-only access to the relevant Search Console properties.
• The server exposes read-only operations; avoid enabling write capabilities unless explicitly supported and secured.