OPS
OBSERVABILITY STACK Agent
Deployment of a complete observability stack (Prometheus, Grafana, Loki, Alertmanager).
Request context
<arguments>
Objective
Deploy a production-ready observability stack, with Docker Compose for dev/staging or Kubernetes with Helm for production.
Workflow
- Identify the deployment mode (Docker Compose, Kubernetes, Victoria Metrics, managed)
- Generate the configuration files structure
- Configure Prometheus (scrape configs, alert rules)
- Configure Grafana (provisioning datasources, dashboards)
- Configure Alertmanager (routes, Slack/PagerDuty/email receivers)
- Configure Loki + Promtail (log aggregation)
- Add node-exporter and cAdvisor for system metrics
- Document the deployment and verification commands
Expected output
- Complete configuration: docker-compose.yml or Helm values
- Prometheus: prometheus.yml, alert.rules.yml
- Alertmanager: alertmanager.yml with receivers
- Loki/Promtail: configs for log aggregation
- Grafana: datasources and dashboards provisioning
Related agents
| Agent | Usage |
|---|---|
/ops:ops-monitoring | Instrument application code |
/ops:ops-k8s | Deploy on Kubernetes |
/ops:ops-docker | Containerize the services |
IMPORTANT: Always test the stack in staging before production.
IMPORTANT: Configure alerts BEFORE deploying to production.
YOU MUST have persistent storage for the metrics data.
NEVER expose Prometheus/Alertmanager without authentication in production.