Official Python SDK for building Model Context Protocol servers and clients that connect LLM applications to tools, data and prompts.
exa-labs/exa-mcp-server
An MCP server that connects AI agents to Exa for web search, page content fetching, and multi-step research.
- Stars
- 5,149
- Forks
- 422
- Open issues
- 20
- Last commit
- 22 hours ago
What is it?
Exa MCP Server exposes Exa's web search and crawling capabilities through the Model Context Protocol, so AI clients such as Claude, Cursor, VS Code, and Codex can call them as tools. The hosted endpoint at https://mcp.exa.ai/mcp can be added to a client directly, and the project also ships as an npm package and as an agent plugin.
By default it provides two tools: web_search_exa, which returns clean, ready-to-use content for a query, and web_fetch_exa, which reads one or more URLs and returns their full content as markdown. Optional tools add advanced search with filters, domains, dates, highlights and subpage crawling, and agent_run for multi-step research and structured output. Tools are selected through a query parameter on the MCP URL.
Anonymous use is rate limited. OAuth or an Exa API key, passed in the URL or in a header, raises the limits and is required for the Exa Agent tool.
Key features
- 01Provides web_search_exa for search with clean content results.
- 02Provides web_fetch_exa to read full page content as markdown from URLs.
- 03Offers optional advanced search with filters, domains, dates and subpage crawling.
- 04Includes agent_run for multi-step research and structured output.
- 05Works with clients such as Claude, Cursor, VS Code, Codex and Gemini CLI.
Use cases
- Giving a coding assistant live web search while it answers questions about current libraries or APIs.
- Letting an AI agent read specific documentation pages or articles by URL.
- Running multi-step competitive analysis or lead research through an agent workflow.
Quick start
- 1
Connect to Exa's hosted MCP server at the URL
- 2
Install for Claude with the plugin command
claude plugin install exa@claude-plugins-official - 3
Install for Codex with the MCP add command
codex mcp add exa --url https://mcp.exa.ai/mcp
Things to know
- Anonymous use is rate limited; higher limits and Exa Agent need OAuth or an API key.
- Exa Agent requires authentication.
Topics
FAQ
- What is exa-mcp-server?
- An MCP server that connects AI agents to Exa for web search, page content fetching, and multi-step research. Exa MCP Server exposes Exa's web search and crawling capabilities through the Model Context Protocol, so AI clients such as Claude, Cursor, VS Code, and Codex can call them as tools. The hosted endpoint at https://mcp.exa.ai/mcp can be added to a client directly, and the project also ships as an npm package and as an agent plugin.
- Is exa-mcp-server free to use? What license does it have?
- exa-mcp-server is open source under the MIT license. Check the license terms before using it in commercial products.
- Is exa-mcp-server still maintained?
- Actively maintained. The last commit to exa-mcp-server was on Oct 19, 2026.
- What can I use exa-mcp-server for?
- Giving a coding assistant live web search while it answers questions about current libraries or APIs. Letting an AI agent read specific documentation pages or articles by URL. Running multi-step competitive analysis or lead research through an agent workflow.
- How do I install exa-mcp-server?
- Connect to Exa's hosted MCP server at the URL Install for Claude with the plugin command: claude plugin install exa@claude-plugins-official Install for Codex with the MCP add command: codex mcp add exa --url https://mcp.exa.ai/mcp
- Where can I get exa-mcp-server?
- The source code and installation instructions are on GitHub: https://github.com/exa-labs/exa-mcp-server
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
An MCP server that gives AI assistants like Claude and Cursor documentation, properties and operations for n8n workflow nodes so they can build and validate workflows.
The official TypeScript SDK for building Model Context Protocol servers and clients that give LLM applications standardized context.
Reference implementations of Model Context Protocol servers giving LLM clients access to files, Git, web content, memory and more.
An MCP server and CLI that feed up-to-date, version-specific library docs and code examples into AI coding agents.
A TypeScript framework for building MCP servers, MCP Apps with React views, and clients, with a built-in Inspector and CLI.