Grafana Agent health check

[saika0721]

Grafana OSS v10.1, Loki, Prometheus has been installed and running properly.
Basic necessary settings (such as datasources, dashboard, alert rules etc) has been complete and working normally.
There are over 1000+ servers (Windows / Linux) installed Grafana agent and reporting metrics/logs to prometheus and Loki
But I hit issues on monitor Grafana agent healthiness.
When Grafana agent stopped / crash unexpectedly due to what ever reason, seems cannot monitor Grafana agent healthiness.
May I know how to monitor agent healthiness in Grafana alert rules?
I have integrations/agent sending to prometheus every 30seconds.
Below query tested but cannot show healthiness....
topk(1, label_replace({job="integrations/agent"}, "state", "not_present", "", ""))
count({job="integrations/agent"} == 0) by (instance) < 0

agent:
enabled: true
relabel_configs:
action: replace
source_labels:
agent_hostname
target_label: instance
action: replace
target_label: job
replacement: “integrations/agent”
metric_relabel_configs: