facts checked on
Two monitoring models
- Uptime and API checks
- PostDeploy sends the request on a schedule: HTTP, keyword, TCP or DNS. Checks run from one AWS region. Failed checks open an incident and notify your alert channels.
- Heartbeats
- Your job sends the request: start, finish or fail. PostDeploy records each finish and fail as a check. A fail opens an incident at once.
Pick the model for each thing you run
- A website, API or health endpoint: an HTTP or keyword check.
- A database port or mail server: a TCP check.
- A DNS record you depend on: a DNS check.
- A cron job, backup, queue worker or scheduled workflow: a heartbeat.
Create a monitor
Ask your connected agent. It creates the monitor through MCP, and the Monitor screen shows the result. The package sets the shortest interval.
Prompt
text
Create an HTTP monitor for https://example.com/health that checks every 5 minutes and alerts by email.Get alerts
Alerts go to email, Slack, Discord, a webhook or Telegram. Ask your agent to add the channel to the project, then trigger a test failure before you rely on it.
Limits
- No ICMP ping, UDP, or browser checks.
- No multi-region confirmation. Checks run from one AWS region.
- A heartbeat that stops arriving does not open an incident or send an alert.