// the find
prometheus-community/helm-charts
Prometheus community Helm charts
The official community home for Prometheus Helm charts, most notably kube-prometheus-stack, which bundles Prometheus Operator, Alertmanager, Grafana, and a pile of pre-built recording rules and dashboards into one chart. If you're deploying monitoring on Kubernetes and not rolling everything by hand, this is where you start.
The kube-prometheus-stack chart is genuinely well-thought-out: CRDs are split into their own sub-chart so you can manage them independently during upgrades, which saves you from the classic 'helm upgrade blew away my CRDs' disaster. Chart signing is in place and the CI pipeline actually runs helm lint and chart-testing against multiple Kubernetes versions, so regressions get caught before release. The bundled Grafana dashboards and Prometheus recording rules are synced from upstream mixins via scripts rather than hand-maintained, which keeps them from rotting. The repo ships real unit tests for Helm templates using helm-unittest, not just CI smoke tests.
kube-prometheus-stack's values.yaml is enormous — over 4,000 lines — and the surface area of what you can configure has grown to the point where upgrades frequently require consulting UPGRADE.md just to avoid breaking changes. The chart still uses job-patch Jobs (with associated RBAC) to manage webhook certs by default instead of cert-manager, which is a fragile pattern that breaks on clusters with strict PodSecurityAdmission. Individual charts in the repo are maintained at very different quality levels; some smaller charts (jiralert, for example) have sparse documentation and no unit tests. There's no good story for GitOps workflows that want to pin CRD versions independently of the main chart version — you either take the bundled CRDs or manage them entirely yourself.