MCP Integration (AI Assistants)

Connect exit1.dev to AI assistants like Claude, Cursor, and Windsurf using the Model Context Protocol (MCP) . Set up monitoring straight from your editor, then ask questions about your monitoring data in natural language. No dashboards required.

Available on every plan, including Free. MCP access follows API access: they are the same entitlement, not two. Key counts: 1 on Free, 3 on Indie, 10 on Nano, 100 on Pro.

There are two ways to connect:

• Hosted (recommended). OAuth sign-in through your browser. No API key, no npm install, and it includes the write tools (create checks, configure alerts, send test alerts).
• Local stdio. The exit1-mcp npm package with an API key. Use it when your client doesn’t support remote MCP servers with OAuth, or in CI-ish setups where a browser step is unwelcome.

Hosted setup (recommended)

Run this in your terminal before starting a session. MCP clients load servers at startup, so a server added mid-conversation isn’t usable until you restart the client.

claude mcp add --transport http exit1 https://app.exit1.dev/mcp/v1 && claude mcp login exit1

The login opens your browser to sign in. No Exit1 account yet? Sign-up happens right there in the flow.

It’s one chained line on purpose: pasting the commands as two separate lines is unreliable in some terminals (cmd.exe drops the second line). Two caveats: in old Windows PowerShell (5.1), where && doesn’t parse, run the two halves one at a time; and on Claude Code versions without claude mcp login, run the add command alone, then start claude and authenticate via /mcp.

Other clients: add https://app.exit1.dev/mcp/v1 as an HTTP MCP server wherever your tool configures MCP, restart it, and complete the sign-in when prompted.

Once connected, paste the setup prompt from app.exit1.dev/mcp  into your assistant from inside your project, or use the built-in setup_monitoring prompt (/exit1:setup_monitoring in Claude Code). The assistant reads your repo, proposes checks, creates them, configures alerts, and sends a test alert.

Local stdio setup (API key)

Requirements

• An Exit1 account on any plan (Free included)
• An API key (add checks:write and alerts:write scopes if the assistant should create checks and configure alerts, not just read)
• Node.js 18+ installed

Step 1: Create an API Key

1. Log in to app.exit1.dev 
2. Go to Settings > API Keys
3. Click Create API Key
4. Name it (e.g. “Claude MCP”)
5. Select the scopes you want the assistant to have
6. Copy the key. You’ll need it in the next step.

Step 2: Connect Your AI Assistant

Claude Desktop

1. Open Claude Desktop
2. Click the menu in the top-left
3. Go to Settings > Developer > Edit Config
4. Add the following to your config:

{
  "mcpServers": {
    "exit1": {
      "command": "npx",
      "args": ["-y", "exit1-mcp"],
      "env": {
        "EXIT1_API_KEY": "ek_live_your_key_here"
      }
    }
  }
}

5. Replace ek_live_your_key_here with your actual API key
6. Save the file and restart Claude Desktop

Claude Code

Run in your terminal:

claude mcp add exit1 -- npx -y exit1-mcp

Then add EXIT1_API_KEY to your shell environment:

# macOS / Linux: add to ~/.bashrc or ~/.zshrc
export EXIT1_API_KEY="ek_live_your_key_here"
 
# Windows: run in PowerShell
[System.Environment]::SetEnvironmentVariable("EXIT1_API_KEY", "ek_live_your_key_here", "User")

Cursor

Add to .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "exit1": {
      "command": "npx",
      "args": ["-y", "exit1-mcp"],
      "env": {
        "EXIT1_API_KEY": "ek_live_your_key_here"
      }
    }
  }
}

Windsurf

Add to your Windsurf MCP config (Settings > MCP):

{
  "mcpServers": {
    "exit1": {
      "command": "npx",
      "args": ["-y", "exit1-mcp"],
      "env": {
        "EXIT1_API_KEY": "ek_live_your_key_here"
      }
    }
  }
}

Start Asking Questions

Once connected, you can ask your AI assistant things like:

• “Set up uptime monitoring for this project”: reads your repo, creates the checks, configures alerts, sends a test alert
• “Are any of my monitors down?”: lists all checks and their status
• “What’s the uptime for my API check over the last 30 days?”: shows uptime %, response times
• “Show me the last 10 failures for production”: fetches recent errors from history
• “Compare response times this week vs last week”: multi-range stats comparison
• “What does my status page show right now?”: current status page snapshot

Available Tools

Both connection methods expose the same 14 tools: 7 read, 7 write.

Rate Limits

The hosted server has its own interactive-session limits (30 requests/min, 200 writes/day). The stdio server uses your API key’s limits:

Normal AI conversations use 2–5 tool calls, so you’re unlikely to hit these limits in typical usage.

Troubleshooting

The assistant can’t see the exit1 tools even though the server is configured Your client loaded its MCP servers before the exit1 entry existed. Restart the client (in Claude Code, claude --continue resumes the conversation with the new server loaded).

“EXIT1_API_KEY environment variable is required” Your API key isn’t set. Double-check the env block in your config and make sure the key is correct.

“API error 401: Unauthorized” Your API key is invalid or has been revoked. Create a new one in the Exit1 dashboard. On the hosted server, run claude mcp login exit1 again (or /mcp in a session) to re-authenticate.

“API error 429: Rate limit exceeded” You’ve hit the API rate limit. Wait a moment and try again.

MCP server not showing up in Claude Desktop Make sure you restarted Claude Desktop after editing the config. Check that Node.js 18+ is installed (node --version).

NPM package not found Run npx -y exit1-mcp manually in your terminal to verify it installs correctly. If behind a corporate proxy, configure npm’s proxy settings.