home / mcp / mcp search server

MCP Search Server

An intelligent MCP (Model Context Protocol) server that helps you discover and research MCP servers using the powerful Exa AI search engine. Built with FastMCP for seamless integration with AI assistants like Claude, Cursor, and more.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "haran2001-mcp-search-server": {
      "url": "https://mcp.example.com/mcp",
      "headers": {
        "EXA_API_KEY": "your_exa_api_key_here"
      }
    }
  }
}

You can run the MCP Search Server locally to discover and research MCP servers using Exa AI, analyze results, and interact with MCP data from your favorite clients. This server is designed to be hosted locally, integrated with AI assistants, and queried to find, compare, and learn about MCPs relevant to your projects.

How to use

You connect to the MCP Search Server from an MCP client or a dev environment. Start the server locally and use the supported MCP tools to perform searches for MCPs based on requirements, fetch detailed MCP information, locate similar MCPs, and ask targeted questions. Review confidence scores and categories to prioritize which MCPs to evaluate first. When you find candidates, retrieve their installation notes and repository details to compare options side by side.

How to install

Prerequisites
- Python 3.10 or higher
- Exa API Key obtained from the Exa Dashboard
- Internet access for search functionality

Installation steps
1. Create a project directory
2. Install dependencies
3. Set the API key
4. Run the server

Install dependencies and start the server exactly as shown in the flow below. You can adapt paths to your environment, but the runtime commands and file names must remain intact.

Configuration and usage notes

Environment variables are required to access Exa services. Set the API key in your environment before starting the server. The key should be named EXA_API_KEY and can be set in your shell or system environment. The server runs by executing the Python script that hosts the MCP services. You can also run it through a virtual environment for isolation.

# Example: running in a virtual environment
python -m venv venv
source venv/bin/activate  # Linux/macOS
venv\Scripts\activate     # Windows
pip install -r requirements.txt
export EXA_API_KEY=your_exa_api_key_here
python mcp_search_server.py

Security and privacy

The server uses the Exa API securely and performs read-only search operations. Your API key is required to access Exa services and is not logged or exposed by the server. Ensure you rotate keys according to your security policies and never share the key in public channels.

Testing and troubleshooting

Verify your environment and dependencies before running. If you encounter issues, confirm that the EXA_API_KEY is set correctly and that Python and required packages are installed. If a network or API issue occurs, check your internet connection and the validity of your API key. Use the provided test script to validate server functionality.

# Basic tests
python --version
python -c "import fastmcp, httpx; print('Dependencies OK')"
export EXA_API_KEY=exa_your_key_here
python mcp_search_server.py

Use cases and capabilities

Leverage the server to discover MCPs that match your needs, get detailed MCP information, compare similar MCPs, and ask questions about MCP capabilities. Use categorized results to explore by functional areas such as database operations, web APIs, file management, and AI/ML integrations. This helps you plan integrations, evaluate deployment options, and speed up decision making.

API and integration notes

You can integrate the MCP Search Server with clients and automation tools using the supported tools exposed by the MCP framework. Look up MCP details, search for MCPs, find similar options, and ask targeted questions to build a knowledge base for your team.

Available tools

search_mcps

Search for MCPs based on requirements with results including confidence scores and categories.

get_mcp_details

Retrieve detailed information about a specific MCP from a URL or repository.

find_similar_mcps

Find MCPs similar to a reference MCP URL with comparison data.

ask_mcp_question

Ask focused questions about MCPs and receive direct answers with citations.

categorize_mcps

Organize MCPs by functional categories based on a given requirement.