cronalert-uptime-monitoring

CronAlert uptime monitoring

CronAlert checks user-supplied URLs and TCP ports on a schedule from Cloudflare's edge, waits for heartbeat pings from cron jobs, opens an incident when something fails, and alerts the team over email, Slack, Discord, webhook, Microsoft Teams, Telegram, PagerDuty, or push. This skill tells you when and how to drive it through the cronalert-mcp tools.

Prerequisites

Check that the cronalert MCP server is connected (its tools are named list_monitors, create_monitor, and so on). If it is not, walk the user through setup before doing anything else:

1. Create a free account at https://cronalert.com if they do not have one.
2. Create an API key at https://cronalert.com/app/settings/api-keys. Recommend a read-only key unless the user explicitly needs you to create, change, or delete monitors. Read-only keys are rejected server-side on every write, so they are the safe default for an agent.
3. Register the server with their client:

# Claude Code
claude mcp add cronalert -e CRONALERT_API_KEY=ca_xxx -- npx -y cronalert-mcp

// Cursor (.cursor/mcp.json), Claude Desktop, Windsurf
{
  "mcpServers": {
    "cronalert": {
      "command": "npx",
      "args": ["-y", "cronalert-mcp"],
      "env": { "CRONALERT_API_KEY": "ca_xxx" }
    }
  }
}

A hosted endpoint also exists at https://cronalert.com/mcp with Authorization: Bearer ca_xxx for clients that support Streamable HTTP.

Never ask the user to paste their API key into chat. Point them at the config file.

Choosing the monitor type

Pick the type before calling create_monitor. Getting this wrong is the most common mistake.

Situation	Type	Notes
Website, API, health endpoint, anything with a URL	http	Default. Checks status code, optionally response content.
Cron job, scheduled task, queue worker, backup script	heartbeat	CronAlert waits for the job to ping it. Alerts when the ping is late. Pro+.
Database, Redis, SMTP (not port 25), SSH, game server, anything on a bare IP	tcp	Opens a connection to host:port. Pro+.