Provides real-time web search via Exa API with structured results, caching, and live crawling for MCP clients.
Configuration
View docs{
"mcpServers": {
"aditya002p-mcp-server": {
"command": "npx",
"args": [
"/path/to/exa-mcp-server/build/index.js"
],
"env": {
"EXA_API_KEY": "your-api-key-here"
}
}
}
}You can run an MCP server that lets AI assistants access Exa’s live web search capabilities in a controlled, real-time way. This guide shows you how to set up the Exa MCP Server, connect it to your MCP client, and use it to perform safe, structured web searches with caching and error handling.
Once you configure your MCP client to connect to the Exa MCP Server, you can ask your AI assistant to perform web searches and retrieve structured results. The server queries Exa’s search API, returns titles, URLs, and content snippets, caches recent searches for future reference, and handles live crawling for fresh content. You interact with it through your MCP client’s normal prompts, without worrying about lower-level API calls.
Prerequisites: install Node.js (v18 or higher), ensure you have Git, and have an Exa API key available. You also need an installation method for your MCP client (Claude Desktop is used in examples). Start by verifying Node.js with this command:
node --version # Should show v18.0.0 or higherChoose an installation approach that matches your setup.
npm install -g exa-mcp-serverIf you use Smithery to integrate Claude Desktop automatically, configure the client with Smithery and the Exa MCP server will be wired up for you.
npx -y @smithery/cli install exa --client claudeFor manual installation, clone the project, install dependencies, build, and link the executable so you can run it from anywhere.
git clone https://github.com/exa-labs/exa-mcp-server.git
cd exa-mcp-server
npm install
npm run build
npm linkConfigure Claude Desktop to recognize the Exa MCP server by adding a server entry that points to the local build index. You will set the command to run npx and pass the path to the server build, along with your API key for Exa.
{
"mcpServers": {
"exa": {
"command": "npx",
"args": ["/path/to/exa-mcp-server/build/index.js"],
"env": {
"EXA_API_KEY": "your-api-key-here"
}
}
}
}If you want to enable specific tools, you can pass a tools list to the server on startup. Tools include web_search_exa, research_paper_search, twitter_search, company_research, crawling, competitor_finder, and linkedin_search.
{
"mcpServers": {
"exa": {
"command": "npx",
"args": [
"/path/to/exa-mcp-server/build/index.js",
"--tools=web_search_exa,research_paper_search,twitter_search,company_research,crawling,competitor_finder,linkedin_search"
],
"env": {
"EXA_API_KEY": "your-api-key-here"
}
}
}
}Test connectivity and capabilities with the MCP Inspector or by starting Claude Desktop and confirming the Exa server connection (the MCP connection is shown with a plugged-in icon). Run the server and verify it accepts queries and returns structured results.
Common issues include invalid API keys, misconfigured client settings, or connection problems. Ensure EXA_API_KEY is valid and correctly set in the client configuration, that Node.js is installed, and that Claude Desktop is restarted after configuration changes.
Keep your API key secure and rotate it if you suspect it has been compromised. Regularly update the MCP server package to receive the latest features and security fixes.
When you ask your AI to search, the server queries Exa’s API, optionally crawls for fresh content, formats results with titles, URLs, and snippets, and caches recent searches for reference.
Performs real-time web searches with optimized results and content extraction.
Specialized search focused on academic papers and research content.
Dedicated Twitter/X.com search for posts, profiles, and conversations.
Crawls company websites to gather detailed business information.
Extracts content from specific URLs, including articles and PDFs.
Identifies competitors by searching for similar products or services.
Search LinkedIn for companies and people using Exa AI.