MCP Server Search

An MCP (Model Context Protocol) server that provides Google search functionality for AI models. This server allows models to search for up-to-date information from the web.

Features

• Google search integration with caching
• Configurable request delays and retries to avoid rate limiting
• Support for both simple and advanced searches (with descriptions)
• Throttling and randomization to be a good web citizen

Installation

Using Docker (Recommended)

1. Build the Docker image:

   docker build -t mcp-server-search .
   

2. Create required directories for persistence:

   mkdir -p ./logs ./cache
   

3. Run the container:

   docker run --rm -i \
     -v "$(pwd)/logs:/app/logs" \
     -v "$(pwd)/cache:/app/cache" \
     mcp-server-search
   

The volumes ensure:

• Logs are persisted to ./logs directory
• Search cache is persisted to ./cache directory
• Both directories are automatically created by Docker if they don't exist

Note: The -i flag is required for MCP protocol communication, and --rm automatically removes the container when it exits.

Cline Integration

To use this MCP server with Cline, add the following configuration to your Cline MCP settings file (located at ~/.config/Code/User/globalStorage/rooveterinaryinc.roo-cline/settings/cline_mcp_settings.json for VS Code):

{
    "mcpServers": {
        "google_search": {
            "command": "docker",
            "args": [
                "run",
                "--rm",
                "-i",
                "mcp-server-search"
            ],
            "disabled": false,
            "alwaysAllow": []
        }
    }
}

This configuration:

• Sets up the search MCP server to run in a Docker container
• Uses the --rm flag to automatically remove the container when it exits
• Uses -i for interactive mode required by the MCP protocol
• Disables the server by default for security (set disabled to false to enable)
• Requires explicit approval for all tool uses (alwaysAllow is empty)

Configuration

The server accepts the following command-line arguments:

• --log-level: Set the logging level (DEBUG, INFO, WARNING, ERROR, CRITICAL). Default: INFO
• --log-file: Path to log file (default: stdout)
• --cache-path: Path to the search cache file (default: cache/google_cache.db)
• --request-delay: Delay between search requests in seconds (default: 5)
• --max-retries: Maximum number of retries for failed searches (default: 3)

Usage

The server exposes the following MCP endpoints:

Tools

• google_search: Search Google and return results. Results are presented in a formatted Markdown structure with titles, URLs, and descriptions (when enabled).

  Parameters:

  • query (string, required): The search query to execute
  • num_results (integer, optional): Number of results to return (min: 1, max: 20, default: 5)
  • use_cache (boolean, optional): Whether to use cached results if available (default: true)
  • include_descriptions (boolean, optional): Whether to include descriptions in results (default: true)

  Features:

  • Automatic request throttling and retry mechanism
  • Random user agent rotation for better request distribution
  • File-based caching with thread-safe access
  • Exponential backoff on rate limiting (HTTP 429)

Prompts

• google_search: Search Google with the given query. Returns results in a conversational format.

  Parameters:

  • query (string, required): The search query to execute
  • num_results (integer, optional): Number of results to return (1-20, default: 5)

About MCP

The Model Context Protocol (MCP) is a protocol for connecting Large Language Models (LLMs) with tools and data sources. Learn more at github.com/modelcontextprotocol.