MCP Server Search

🚀 MCP Server Search

This is an MCP (Model Context Protocol) server that provides Google search functionality. The server allows AI models to obtain the latest information from the web.

✨ Features

• Integrates Google search with caching functionality.
• Configurable request delay and retry to avoid rate limiting.
• Supports simple search and advanced search (with descriptions).
• Limits traffic and randomizes requests to be a good network citizen.

📦 Installation

Using Docker (Recommended)

1. Build the Docker image:

   docker build -t mcp-server-search .
   

2. Create 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
   

These volumes ensure:

• Logs are persisted to the ./logs directory.
• Search cache is persisted to the ./cache directory.
• Docker will automatically create the directories if they don't exist.

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

Cline Integration

To use this MCP server with Cline, add the following configuration to Cline's 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 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 (set disabled to false to enable).
• Requires explicit approval for all tool usage (alwaysAllow is empty).

🔧 Configuration

The server accepts the following command-line parameters:

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

💻 Usage Examples

Basic Usage

The server exposes the following MCP endpoints:

Tool

• google_search: Performs a Google search and returns the results. Results are presented in Markdown format with titles, URLs, and descriptions (if enabled).

  Parameters:

  • query (string, required): The search query to execute.
  • num_results (integer, optional): The number of results to return (minimum: 1, maximum: 20, default: 5).
  • use_cache (boolean, optional): Whether to use available results from the cache (default: true).
  • include_descriptions (boolean, optional): Whether to include descriptions (default: true).

  Features:

  • Automatic request throttling and retry mechanism.
  • User agent randomization for better request distribution.
  • Cache access with a file system.
  • Traffic limitation.

Advanced Usage

When running in the mcp-server-search container, you can pass parameters via the following command:

docker run --rm -i \
    -v "$(pwd)/logs:/app/logs" \
    -v "$(pwd)/cache:/app/cache" \
    mcp-server-search \
    --log-level INFO \
    --log-file /app/logs/search.log \
    --cache-path /app/cache/google_cache.db \
    --request-delay 2 \
    --max-retries 3

Configuration File

By default, the server loads configuration from config.json. Here is an example:

{
    "log_level": "INFO",
    "log_file": "./logs/search.log",
    // Other configuration items...
}