exa
AI-powered code search MCP server using Exa API for intelligent code search and retrieval in AI assistants
claude mcp add --transport stdio it-beard-exa-server node /path/to/exa-server/build/index.js \ --env EXA_API_KEY="your-api-key-here"
How to use
This MCP server, Exa, exposes AI-powered code search capabilities via the Exa API. It accepts natural language queries and returns relevant code examples and documentation in a structured JSON format, with configurable numbers of results and rich metadata. The server is accessed through the exa:// scheme, with tools and resources that help you perform searches and retrieve results. To use it, configure the MCP settings with your Exa API key, then start the server. You can then issue a search via the exa://search/{query} resource, or retrieve the most recent results through exa://last-search/result. The search tool accepts a JSON input schema containing the query and the number of results you want, enabling you to tailor the breadth of results to your needs.
How to install
Prerequisites:
- Node.js and npm installed on your machine
- Access to the repository code for Exa MCP Server
Installation steps:
-
Clone the repository: git clone https://github.com/it-beard/exo-server.git cd exa-server
-
Install dependencies: npm install
-
Build the project: npm run build
-
Configure your Exa API key in the MCP settings file. Example configuration: { "mcpServers": { "exa": { "command": "node", "args": ["/path/to/exa-server/build/index.js"], "env": { "EXA_API_KEY": "your-api-key-here" } } } }
-
Start the MCP server with your preferred runner (depends on your MCP host). Ensure the environment variable for the API key is available to the process.
Additional notes
Tips and common issues:
- Ensure EXA_API_KEY is correctly set in the MCP configuration and is accessible to the process running index.js.
- The path in args should point to the built index.js file after npm run build.
- If you encounter authentication or quota errors from Exa, verify your API key permissions and limits with the Exa service.
- You can adjust the number of search results via the input schema (numResults field) when calling the exa://search endpoint.
- For troubleshooting, check that Node.js version matches the project requirements and that the build artifacts exist at the specified path.
Related MCP Servers
exa
Exa MCP for web search and web crawling!
Matryoshka
MCP server for token-efficient large document analysis via the use of REPL state
ummon
The semantic layer for software engineering: Connect code to meaning, build on understanding
tavily
AI-powered search capabilities for AI assistants using the Tavily API and Model Context Protocol (MCP)
mcpplay
The FastAPI /docs experience, for MCP servers.
cursor-cortex
Structured memory system for AI assistants. Eliminates context loss with branch notes, tacit knowledge, and project context. Local MCP integration for Cursor IDE.