Kubernetes-native continuous delivery

End-to-end delivery, without the pipeline.

Schedule windows, manual approvals, canaries, environment promotion, alert checks, smoke tests. Everyone can describe the pipeline they want; no tool has ever made it easy to build. Kuberik does, one piece at a time.

Get started Star on GitHub

Apache 2.0 · composes with Flux, OpenKruise, Prometheus, Datadog

How it works

A reactive loop with five clear phases.

Each phase is a Kubernetes resource. Two are yours to compose; three are built in. Nothing else is hidden behind the curtain.

phase 01 release

release

guard

gates

update

health

01 release

Which versions are available?

Flux ImagePolicyOCI pullcosign-verified

02 guard built-in · no config

Is this environment safe to deploy to?

healthynot pausedidleno override

03 gates open class · you own this

Which version can we deploy right now?

schedulemanualafter(env)custom CRD

04 update

Roll out the new version.

Flux KustomizationOpenKruisecanary 0→100%

05 health open class · you own this

Did it deploy successfully?

kstatusmetricsalertssmoke testcustom CRD

composing environments

Each environment is a loop. Compose them with gates.

No central pipeline orchestrates anything. Each environment runs its own loop and opens its gate when conditions are met — including after(env).

dev on commit

after(env)

staging after dev

after(env)

production after staging + schedule

The spec

This is everything you write.

Three Flux resources, one Kuberik resource, one $VERSION placeholder.

fluxapiVersion: image.toolkit.fluxcd.io/v1beta2
kind: ImageRepository
metadata:
  name: my-app
spec:
  image: ghcr.io/me/my-app
---
kind: ImagePolicy
spec:
  policy:
    semver:
      range: ">=0.1.0"

kuberikapiVersion: kuberik.com/v1alpha1
kind: Rollout
metadata:
  name: my-app
spec:
  releasesImagePolicy:
    name: my-app
  bakeDuration: 10m

fluxapiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
  name: my-app
  annotations:
    rollout.kuberik.com/
      substitute.VERSION.from: my-app
spec:
  sourceRef:
    kind: GitRepository
    name: my-app

yoursapiVersion: apps/v1
kind: Deployment
metadata: { name: my-app }
spec:
  template:
    spec:
      containers:
        - name: app
          image: ghcr.io/me/my-app:${VERSION}

One dashboard

Every rollout. Every cluster.

No hub cluster. No stored cross-cluster credentials. RBAC on your own OIDC.

Apps — multi-env spread, version timelines, failing apps surfaced

Rollout overview — deployment pipeline, health checks, resources

Activity — deployment timeline grouped by recency, failures highlighted

Try it

Stop writing rollout glue.
Declare gates and health. Watch the loop run.

Get started Source

$ helm repo add kuberik https://charts.kuberik.com
$ helm install kuberik kuberik/rollout-controller