Metasearch

Metasearch is an MCP server that connects AI assistants to the Tavily API to run web queries. Built for researchers, developers, and users working in Model Context Protocol environments like Claude Desktop or Cursor, it brings real-time search functionality directly into assistant conversations. Instead of switching back and forth between a web browser and an AI interface, users can instruct their assistant to look up current information, local events, or technical documentation. The server receives structured queries, forwards them to the Tavily search service, and parses the output back into the conversation in standard text format. Results include AI-generated summaries, source URIs, and webpage titles, giving language models sufficient context to cite sources accurately. The underlying tool supports both basic and advanced search depths depending on how comprehensive the information retrieval needs to be. It can be run locally using Python and uv or deployed via Docker Compose on systems where desktop clients are unavailable.

Category: Web Search & Research

Tags: metasearch, search, tavily, web search

Visit Metasearch

How to install and configure Metasearch

1. Clone the repository: bash git clone https://github.com/YeonwooSung/metasearch-mcp.git 2. Open your Claude Desktop configuration file (claude_desktop_config.json). 3. Add the server entry under mcpServers: json "mcpServers": { "tavily-search": { "command": "uv", "args": [ "--directory", "/path/to/mcp-server-tavily", "run", "tavily-search" ], "env": { "TAVILY_API_KEY": "YOUR_TAVILY_API_KEY", "PYTHONIOENCODING": "utf-8" } } } 4. Restart Claude Desktop. Alternatively, run docker compose up -d for containerized environments.

What you can do with Metasearch

• Querying real-time event schedules, regional news, or current local announcements directly inside Claude Desktop chats. * Retrieving external technical references and official documentation during coding sessions inside the Cursor Composer Agent. * Conducting in-depth web investigations with Tavily advanced search depth to extract source titles and direct website links. * Operating automated search pipelines in headless environments using the Docker Compose setup and the included Python client.

Key facts

• https://github.com/YeonwooSung/metasearch-mcp
• Web Search & Research
• metasearch, search, tavily, web search

Part of MCP Servers

How do I install Metasearch?

Clone the repository using git clone, then add the server configuration to your Claude Desktop configuration file using uv to run the tavily-search command. You must provide a valid Tavily API key and set the Python encoding to utf-8 in the environment settings before restarting the client.

What can Metasearch do?

Metasearch exposes a search tool to MCP clients that queries the Tavily API. It accepts a search query string and an optional search depth parameter, returning structured search responses that include AI summaries, webpage titles, and source URLs.

Which MCP clients work with Metasearch?

Metasearch works directly with Claude Desktop on Windows and macOS. It can also be configured to work with Cursor using a shell script wrapper, or executed through a local client script inside Docker Compose on other platforms.

What parameters does the search tool require?

The search tool requires a query parameter containing the text string you want to look up. It also accepts an optional search_depth argument, which can be configured to either basic or advanced depending on how thorough the search should be.