Define an SLO with Sloth and generate Prometheus recording and alerting rules

Steps

1. Install the Sloth CLI binary from the project releases page; Sloth reads a custom YAML SLO specification and outputs a Prometheus rules YAML file compatible with the rule_files directive or the Prometheus Operator's PrometheusRule CRD.
2. Write a Sloth SLO spec file with the version, service, labels, and slos fields; each entry in slos requires an objective.target (e.g., 99.5), objective.window (e.g., 30d), sli.events.error_query and sli.events.total_query in PromQL.
3. Run sloth generate -i slo.yaml -o rules.yaml to produce a rules file; the output includes recording rules for all standard SLI windows (5m, 30m, 1h, 2h, 6h, 1d, 3d) and multi-window multi-burn-rate alert rules.
4. Apply the generated rules file to Prometheus by placing it in the rule_files path and reloading, or apply the PrometheusRule CRD manifest to a Kubernetes cluster running the Prometheus Operator.
5. Optionally use the Sloth Kubernetes controller (sloth controller) to watch PrometheusRule CRDs annotated with sloth.dev/mode: controller and generate rules in-cluster without the CLI step.
6. Import the Sloth-provided Grafana dashboard (available in the project repository) to visualize SLO error budget burn rates, remaining budget, and alert status across all defined SLOs.