prometheus.yml lists scrape jobs: where to pull metrics from and how often. Rule files hold two kinds of rule: alerts (an expression that, true for a for: duration, fires) and recording rules (a query saved under a new name).

Prometheus only *detects*. Alertmanager groups, silences and routes alerts to receivers (chat, email, a pager) by their labels.

Good alerts describe symptoms users feel (errors, latency), and page a human only when action is needed now. Nothing here runs a monitoring server: you write the real files and the checks read them the way the tools would.

Prometheus: scrape config and alerts