searchapi

This MCP server exposes an integration point called SearchAPI.site that lets AI assistants query external data sources (such as Google and Bing) via the SearchAPI.site API. It supports a stdio transport for CLI usage and a streamable HTTP transport for web-based clients. Once running, you can invoke the included MCP tools and CLI commands to search the web, images, and videos through the SearchAPI service, enabling your assistants to fetch up-to-date results from supported sources. The server is configured to work with a local development setup by default, and can be exposed remotely using the provided HTTP transport configuration.

Usage paths:

• Local development with stdio transport: use CLI or internal MCP tooling to run searches directly from the command line (e.g., search-google, search-google-images, search-youtube) via npm scripts in development mode.
• Remote HTTP access: configure an HTTP transport to expose an endpoint (default http://mcp.host/mcp) so clients can query the MCP server over HTTP. This is useful for web clients or remote AI agents that rely on HTTP-based MCP calls.
• API surface: the server wires SearchAPI.site endpoints (Google/Bing sources and more as enabled) into MCP tools, enabling your LLMs to perform web searches, image searches, and YouTube search within a controlled MCP context.

Prerequisites:

• Node.js (≥18.x)
• Git

Step-by-step installation:

1. Clone the repository: git clone https://github.com/mrgoonie/searchapi-mcp-server.git cd searchapi-mcp-server

2. Install dependencies: npm install

3. Run the development server (stdio transport): npm run dev:server

4. (Optional) Run the development server with HTTP transport: npm run dev:server:http

5. If you need to run CLI tools in development: npm run dev:cli -- [command] [args]

Notes:

• For production deployments, build the project and run using start scripts as documented in package.json.

• The MCP server supports multiple transports; stdio is the default for CLI usage, and Streamable HTTP is available for web clients. You can also configure an HTTP transport by providing a URL instead of the stdio transport.
• Environment variables for HTTP transport include MCP_HTTP_HOST, MCP_HTTP_PORT, and MCP_HTTP_PATH to customize binding, port, and API path.
• The included CLI commands (search-google, search-google-images, search-youtube) facilitate representative searches against Google/YouTube/Bing via SearchAPI.site; you can adapt or extend tools in the repo to cover additional sources listed in the Available platforms.
• When debugging locally, the MCP Inspector is exposed (e.g., http://127.0.0.1:6274) to inspect MCP traffic and responses.
• If you plan to deploy to production, consider using the http transport with proper authentication and rate limiting, and keep your API keys secure (do not hard-code keys in code or config).