Serper

You run the Serper MCP Server to enable your LLMs to fetch live Google search results via Serper. This server acts as a bridge that translates your MCP client’s requests into Serper API calls and returns structured results for downstream thinking and reasoning.

How to use

Use the Serper MCP Server with an MCP client to access Google search results through Serper. You can run the server locally and connect your client to it either via a standard input/output (stdio) channel or through a managed runtime like uvx. Start the server, configure your client to point to the MCP server, and supply your Serper API key to authorize requests.

How to install

Prerequisites you need on your system before starting:

• Python 3.x installed
• pip (Python package manager)
• uvx tool if you choose stdio via uvx

Choose one of the supported installation approaches below and follow the steps exactly as shown.

How to install

# Option A: Install and run via uvx (stdio) using the Serper MCP Server
# This requires uvx and the Serper MCP Server to be available
npx -y @smithery/cli install @garylab/serper-mcp-server --client claude

{
  "mcpServers": {
    "serper": {
      "command": "uvx",
      "args": ["serper-mcp-server"],
      "env": {
        "SERPER_API_KEY": "<Your Serper API key>"
      }
    }
  }
}

If you prefer direct Python execution, use the Python module form shown here.

# Option B: Use the Serper MCP Server with Python module (stdio)
python3 -m serper_mcp_server
# When running in this mode, set the environment variable for authentication
export SERPER_API_KEY="<Your Serper API key>"

{
  "mcpServers": {
    "serper": {
      "command": "python3",
      "args": ["-m", "serper_mcp_server"],
      "env": {
        "SERPER_API_KEY": "<Your Serper API key>"
      }
    }
  }
}

Additional notes

Remember to provide your Serper API key where indicated. The server will accept connections from your MCP client and translate queries to Serper, returning search results suitable for your LLM workflows.

Security and troubleshooting

securely store your Serper API key and limit access to the MCP server instance. If you encounter connection issues, verify that the MCP client is configured to point to the correct stdio channel or URL and that the API key is valid.

Examples and practical usage

Use the server to answer questions that require up-to-date information from Google, such as current events, product availability, or scholarly references. Your MCP client can request search results and pass them to the LLM for contextual reasoning.

Available tools

google_search

Set all the parameters for a Google search and return the results to the MCP client.

google_search_images

Search for images using Google and return results to the MCP client.

google_search_videos

Search for videos using Google and return results to the MCP client.

google_search_places

Search for places using Google Places and return results.

google_search_maps

Perform Google Maps queries and return map results.

google_search_reviews

Retrieve reviews from Google-related sources and present to the MCP client.

google_search_news

Fetch news results from Google News for the MCP client.

google_search_shopping

Search Google Shopping results and return product listings.

google_search_lens

Query Google Lens-related insights and return results.

google_search_scholar

Query Google Scholar and return academic results.

google_search_patents

Query Google Patents for patent information.

google_search_autocomplete

Get autocomplete suggestions from Google for a given query.

webpage_scrape

Scrape a webpage and return structured data to the MCP client.