Exa MCP connects AI assistants to Exa’s search capabilities, including web search, code search, Exa Agent, and Exa Connect. It is open-source and available on GitHub.

Installation

Exa’s Search MCP can be installed in any MCP client with the server URL: https://mcp.exa.ai/mcp

Install in Cursor

Exa MCP is available on Cursor.

Install in VS Code

Exa MCP is available on VSCode.

Run in terminal:

codex mcp add exa --url https://mcp.exa.ai/mcp

Install the Exa plugin — it includes the MCP server plus Exa’s skills. Run in terminal:

claude plugin install exa@claude-plugins-official

Or in Claude Code, type /plugin, search for Exa, and install it.To add just the MCP server:

claude mcp add --transport http exa https://mcp.exa.ai/mcp

Add to your opencode.json:

{
  "mcp": {
    "exa": {
      "type": "remote",
      "url": "https://mcp.exa.ai/mcp",
      "enabled": true
    }
  }
}

Add to ~/.kiro/settings/mcp.json:

{
  "mcpServers": {
    "exa": {
      "url": "https://mcp.exa.ai/mcp"
    }
  }
}

Exa MCP works with most other MCP clients — point them at https://mcp.exa.ai/mcp. The config key for the URL varies by client:

Client	Where to add it	URL key
Windsurf	`~/.codeium/windsurf/mcp_config.json` (under `mcpServers`)	`serverUrl`
Google Antigravity	Agent panel → Manage MCP Servers → View Raw config (under `mcpServers`)	`serverUrl`
Zed	Zed `settings.json` (under `context_servers`)	`url`
Gemini CLI	`~/.gemini/settings.json` (under `mcpServers`)	`httpUrl`
Warp	Settings → MCP Servers → Add MCP Server (top-level `exa`)	`url`
v0 by Vercel	Prompt Tools → Add MCP	paste the URL directly

Most other clients use the standard mcpServers shape:

{
  "mcpServers": {
    "exa": {
      "url": "https://mcp.exa.ai/mcp"
    }
  }
}

If your client doesn’t support remote MCP servers directly, use the mcp-remote bridge:

{
  "mcpServers": {
    "exa": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.exa.ai/mcp"]
    }
  }
}

Or run the local npm package with your Exa API key:

{
  "mcpServers": {
    "exa": {
      "command": "npx",
      "args": ["-y", "exa-mcp-server"],
      "env": {
        "EXA_API_KEY": "your_api_key"
      }
    }
  }
}

API Key

Get your Exa API key

Exa MCP has a generous free plan. To overcome free plan rate limits and enable production use, add your own API key:

{
  "exa": {
    "url": "https://mcp.exa.ai/mcp",
    "headers": {
      "x-api-key": "YOUR_EXA_API_KEY"
    }
  }
}

Available Tools

Enabled by default:

Tool	Description
`web_search_exa`	Search the web for any topic and get clean, ready-to-use content
`web_fetch_exa`	Read a webpage’s full content as clean markdown from one or more URLs

Additional tools (enable via the tools parameter):

Tool	Description
`web_search_advanced_exa`	Advanced web search with full control over category filters, domain restrictions, date ranges, highlights, summaries, and subpage crawling
`agent_tools`	Alias that enables the Exa Agent tool for multi-step research, list-building, and enrichment

Enable specific tools:

https://mcp.exa.ai/mcp?tools=web_fetch_exa

Enable all search tools (see Exa Agent below to add the Agent tool):

{
  "exa": {
    "url": "https://mcp.exa.ai/mcp?tools=web_search_exa,web_fetch_exa,web_search_advanced_exa",
    "headers": {
      "x-api-key": "YOUR_EXA_API_KEY"
    }
  }
}

Exa Agent

Exa Agent is now available within Exa MCP. Use it for multi-step research, list-building, enrichment, and structured output when a task needs more than a single search call. Agent runs are usage-based, so the Agent tool requires authentication — connect with OAuth or pass your own Exa API key. Enable the Agent tool with the agent_tools alias:

https://mcp.exa.ai/mcp?tools=agent_tools

Or alongside the default search tools:

{
  "exa": {
    "url": "https://mcp.exa.ai/mcp?tools=web_search_exa,web_fetch_exa,agent_tools",
    "headers": {
      "x-api-key": "YOUR_EXA_API_KEY"
    }
  }
}

The Agent tool

Tool	Description
`agent_run`	Run an Exa Agent end-to-end in a single call. Streams the run to completion and returns the final output — text, structured JSON, grounding citations, usage, and cost.

How it works

agent_run runs the whole agent loop in one call: it creates the run, streams its lifecycle to completion, and returns the final output.

Run the agent with agent_run, passing a natural-language query. Include an outputSchema whenever you need repeatable, structured results.
Read the output. When the run finishes, the response has outputReady: true with output.text, output.structured (when a schema was provided), and output.grounding citations, plus usage and cost.
Long runs. If a run outlives the call window (~750s), agent_run returns status: "running" with the run’s id instead of an error — the run keeps executing server-side. Call agent_run again with runId set to that id to keep waiting until it finishes.
Continue. Pass previousRunId (a completed run’s id) to agent_run to refine or extend earlier work; use input.exclusion to avoid resurfacing prior results.

agent_run accepts a natural-language query (or a runId to resume waiting on a retained run, or a previousRunId to continue completed work) plus optional outputSchema, systemPrompt, input (data to enrich, exclusion to avoid), dataSources (Exa Connect providers, up to 5), and effort (minimal, low, medium, high, xhigh, or auto; defaults to low). See the Exa Agent guide for schema patterns, effort modes, Exa Connect data sources, and pricing.

Resources

GitHub - View Exa MCP source code
npm - Install Exa MCP npm package

Usage Examples

Web Search

Search for recent developments in AI agents and summarize the key trends.

Code Search

Find Python examples for implementing OAuth 2.0 authentication.

Read a Page

Fetch the full content of https://exa.ai and summarize what the company does.

Troubleshooting

Rate limit error (429)You’ve hit the free plan rate limit. Add your own API key to continue:

{
  "exa": {
    "url": "https://mcp.exa.ai/mcp",
    "headers": {
      "x-api-key": "YOUR_EXA_API_KEY"
    }
  }
}

Get your API keyTools not appearingRestart your MCP client after updating the config file. Some clients require a full restart to detect new MCP servers.Claude Desktop not connectingUse the built-in Connector: click + (or Add connectors) → Connectors tab → search for Exa → click +.Config file not foundCommon config locations:

Cursor: ~/.cursor/mcp.json
VS Code: .vscode/mcp.json (in project root)
Claude Desktop (macOS): ~/Library/Application Support/Claude/claude_desktop_config.json
Claude Desktop (Windows): %APPDATA%\Claude\claude_desktop_config.json

API

Data

MCP

SDKs

Integrations

Admin

Additional Resources

Web Search MCP

Installation

Install in Cursor

Install in VS Code

API Key

Get your Exa API key

Available Tools

Exa Agent

The Agent tool

How it works

Resources

​Installation

Install in Cursor

Install in VS Code

​API Key

Get your Exa API key

​Available Tools

​Exa Agent

​The Agent tool

​How it works

​Resources

Installation

API Key

Available Tools

Exa Agent

The Agent tool

How it works

Resources