Exa

You can use the Exa MCP Server to empower AI assistants with real-time web information from the Exa AI Search API. This server lets Claude and other clients query web data safely and at scale, exposing a set of tools that perform live searches, research, and content extraction.

How to use

Connect to the Exa MCP Server from your MCP client or Claude Desktop to start making live web queries. You can use the remote Exa MCP endpoint or run a local MCP server with your API key. The remote endpoint requires your Exa API key and can be configured in your client to enable tools like web search and company research. If you prefer running locally, start the local MCP server and point your client to the local host while providing your API key.

How to install

Prerequisites: ensure you have Node.js and npm installed on your system. You will need an Exa API key with access to the MCP features.

Remote MCP (live Exa hosted server) can be used directly by configuring your MCP client to hit the Exa endpoint and supplying your API key.

To run the Exa MCP Server locally, install the package globally and start it with your API key set in the environment.

npm install -g exa-mcp-server

# Run locally with your API key
EXA_API_KEY=your-api-key-here npx exa-mcp-server

Additional setup and notes

Configure the client to recognize the Exa MCP server by adding it as an MCP server with the appropriate command and environment variables. You can either use the remote URL directly in your client configuration or run the server locally and point the client to the local MCP endpoint.

Available tools

web_search_exa

Performs real-time web searches with optimized results and content extraction.

company_research

Crawls company websites to gather detailed information about businesses.

crawling

Extracts content from specific URLs, useful for reading articles and PDFs.

linkedin_search

Search LinkedIn for companies and people using Exa AI.

deep_researcher_start

Starts a smart AI researcher to explore complex questions and produce a detailed research report.

deep_researcher_check

Checks whether an ongoing research task is complete and retrieves the full report.