MCP Server
MCP Server
The WatchCron MCP server lets AI assistants manage your monitors through natural language. It implements the Model Context Protocol (MCP) — an open standard that connects AI agents to external tools. Instead of writing API calls, your AI assistant can create monitors, check statuses, manage incidents, and more.
Supported clients
Any MCP-compatible AI client works with the WatchCron server. Tested with:
- Claude Desktop — Anthropic's desktop app
- Claude Code — CLI agent for software engineering
- Cursor — AI-powered code editor
- Windsurf — AI IDE by Codeium
- ChatGPT (with MCP plugin support)
- Any client that supports the MCP stdio transport
Installation
The MCP server is distributed as an npm package. No global installation required — use npx to run it directly:
npx watchcron-mcp
Or install globally:
npm install -g watchcron-mcp
Configuration
The MCP server needs your WatchCron API key. Pass it as an environment variable:
Claude Desktop
Add to your claude_desktop_config.json:
{
"mcpServers": {
"watchcron": {
"command": "npx",
"args": ["watchcron-mcp"],
"env": {
"WATCHCRON_API_KEY": "YOUR_API_KEY"
}
}
}
}
Claude Code
claude mcp add watchcron -- npx watchcron-mcp
Then set the environment variable in your shell profile or .env file:
export WATCHCRON_API_KEY="YOUR_API_KEY"
Cursor
Add to your .cursor/mcp.json in the project root:
{
"mcpServers": {
"watchcron": {
"command": "npx",
"args": ["watchcron-mcp"],
"env": {
"WATCHCRON_API_KEY": "YOUR_API_KEY"
}
}
}
}
Windsurf
Add to your ~/.codeium/windsurf/mcp_config.json:
{
"mcpServers": {
"watchcron": {
"command": "npx",
"args": ["watchcron-mcp"],
"env": {
"WATCHCRON_API_KEY": "YOUR_API_KEY"
}
}
}
}
Available tools (43)
The MCP server exposes 43 tools covering all 9 API resource types. Each tool maps to one REST API endpoint. Your AI assistant discovers these tools automatically.
Cron Job Monitors (Checks)
| Tool | Description |
|---|---|
list_monitors | List all cron job checks with optional tag filter |
create_monitor | Create a cron job check (simple or cron schedule) |
get_monitor | Get check details by UUID |
update_monitor | Update check settings |
delete_monitor | Delete a check and all ping data |
pause_monitor | Pause or resume a check |
get_pings | List ping history for a check |
HTTP / Uptime Monitors
| Tool | Description |
|---|---|
list_http_monitors | List all HTTP monitors with optional status filter |
create_http_monitor | Create an HTTP monitor (URL, keywords, SSL, thresholds) |
get_http_monitor | Get HTTP monitor details |
update_http_monitor | Update HTTP monitor settings |
delete_http_monitor | Delete an HTTP monitor |
pause_http_monitor | Pause or resume an HTTP monitor |
get_http_monitor_logs | List check logs (status, response time, errors) |
Cloud Cron Jobs
| Tool | Description |
|---|---|
list_cloud_crons | List all Cloud Cron jobs |
create_cloud_cron | Create a scheduled HTTP job (cron expression, retries, overlap prevention) |
get_cloud_cron | Get Cloud Cron job details |
update_cloud_cron | Update a Cloud Cron job |
delete_cloud_cron | Delete a Cloud Cron job |
pause_cloud_cron | Pause or resume a Cloud Cron job |
get_cloud_cron_logs | List execution logs |
Port Monitors (TCP/UDP/DNS/ICMP)
| Tool | Description |
|---|---|
list_port_monitors | List all port monitors with optional type filter |
create_port_monitor | Create a port monitor (TCP, UDP, DNS, or ICMP) |
get_port_monitor | Get port monitor details |
delete_port_monitor | Delete a port monitor |
pause_port_monitor | Pause or resume a port monitor |
Domain Monitors
| Tool | Description |
|---|---|
list_domain_monitors | List all domain expiration monitors |
create_domain_monitor | Create a domain monitor with warning threshold |
get_domain_monitor | Get domain monitor details (registrar, expiry date) |
delete_domain_monitor | Delete a domain monitor |
Blocklist Monitors
| Tool | Description |
|---|---|
list_blocklist_monitors | List all blocklist monitors |
create_blocklist_monitor | Create a DNSBL monitor for an IP or hostname |
get_blocklist_monitor | Get blocklist monitor details |
delete_blocklist_monitor | Delete a blocklist monitor |
list_incidents_log | List blocklist scan logs |
Incidents
| Tool | Description |
|---|---|
get_incidents | List incidents with status and active_only filters |
create_incident | Create an incident with title, severity, and initial message |
get_incident | Get incident details with update timeline |
update_incident | Update status, add timeline message, resolve |
delete_incident | Delete an incident |
Status Pages & Channels
| Tool | Description |
|---|---|
list_status_pages | List all status pages |
get_status_page | Get status page details by slug |
list_channels | List notification channels |
Usage examples
Once configured, you can ask your AI assistant to manage monitors in natural language:
# Create a cron job monitor "Create a check for my nightly backup that runs at 2 AM with a 10-minute grace period" # Set up uptime monitoring "Monitor https://api.example.com/health every minute, alert if it returns anything other than 200" # Schedule a Cloud Cron job "Set up a Cloud Cron to call my WordPress wp-cron.php every 5 minutes" # Check port connectivity "Create a TCP monitor for my database at db.example.com:5432" # Monitor domain expiration "Track when example.com expires and warn me 60 days before" # Manage incidents "Create a critical incident titled 'Database outage' with a message about connection failures" # Check status "Show me all monitors that are currently down" "What's the uptime log for my production API monitor?"
How it works
The MCP server runs locally on your machine as a subprocess. When your AI assistant needs to interact with WatchCron, it:
- Discovers the available tools from the MCP server
- Decides which tool to call based on your request
- Sends structured parameters to the MCP server via stdio
- The MCP server calls the WatchCron REST API with your API key
- Results are returned to your AI assistant in a readable format
No data leaves your machine except the API calls to watchcron.com. The MCP server has no database, no state, and no background processes. It only runs when your AI client invokes it.
Troubleshooting
"Tool not found" or server not connecting
Make sure you have Node.js 18+ installed. Check that the config JSON path is correct for your client. Restart the AI client after changing the config.
Authentication errors
Verify your WATCHCRON_API_KEY is set correctly in the env block. The key must be a valid project-scoped API key from your dashboard. Test it with curl:
curl -H "Authorization: Bearer YOUR_API_KEY" \ https://watchcron.com/api/v1/checks
Rate limiting
The MCP server uses the same rate limits as the REST API. If your AI agent makes too many calls in quick succession, requests may return 429 errors. See Rate limits for details per plan.
Links
- npm: npmjs.com/package/watchcron-mcp
- GitHub: github.com/watchcron/watchcron-mcp
- API docs: REST API Overview
- MCP spec: modelcontextprotocol.io