The Serper MCP Server exposes Google search results to your LLMs via the Serper API. It provides a collection of search tools that you can configure in your MCP client, including general web search, image search, video search, news, maps, places, reviews, shopping, and more. Each tool is designed to mirror a corresponding Serper API feature and accepts a shared set of parameters defined in the server's schemas, allowing you to tailor queries and facets according to your use case. By running the server with an API key, you can route your model queries through the MCP layer to retrieve structured search results that your agent can consume and reason about.

Prerequisites:

• Python 3.11+
• pip
• Serper API key

Installation steps:

1. Install uv (if not already installed) and the Serper MCP server package

Install uv (example for macOS/Linux; Windows commands may differ)

python -m pip install uvx

Serper MCP Server is published as a Python package; you can install it via PyPI when needed

python -m pip install serper-mcp-server

2. Configure the MCP client to use the server

In your MCP client config (Claude or other client), add: { "mcpServers": { "serper": { "command": "uvx", "args": ["serper-mcp-server"], "env": { "SERPER_API_KEY": "<Your Serper API key>" } } } }

3. Run the server via uvx

uvx serper-mcp-server

4. Alternative: Running as a Python module (if you prefer direct Python execution)

python -m serper_mcp_server

Ensure you pass the API key via environment variable or config as above

Notes:

• Always provide your Serper API key in the MCP server configuration (ENV VAR SERPER_API_KEY).
• The server exposes many tools; you can enable specific ones by configuring your MCP client to call the desired tool names (e.g., google_search, google_search_images, webpages_scrape, etc.).
• When debugging with the MCP inspector, you can run: npx @modelcontextprotocol/inspector uvx serper-mcp-server to validate configuration and responses.
• If you upgrade the Serper MCP Server package, re-check the parameter names in src/serper_mcp_server/schemas.py to ensure compatibility with your client calls.