Run Kubernetes with context, in one place
Moonin turns your cluster events into a clear operational view: what changed, what is affected and what to do next.
One Helm chart per cluster. Distributed traces with eBPF, no code instrumentation. Works with EKS on AWS, GKE on GCP, AKS on Azure and self-managed clusters. Available since August 2026.
The trace does not live in your code
eBPF watches traffic from the node kernel. No SDK to add, no build to repeat, no language left out: the containers stay untouched and the observation happens underneath them.
Startup programs
We are part of the Google Cloud, AWS and Microsoft startup programs: Google for Startups Cloud Program, AWS Activate and Microsoft for Startups.
Frequently asked questions
What is Moonin?
Moonin is an operational observability platform for Kubernetes, built by Arguz SpA in Santiago, Chile. It correlates a cluster’s resources, deployments and events to answer four questions during an incident: what changed, what is affected, who needs to know, and what to do next. It installs as a single Helm chart per cluster.
Is Moonin a Kubernetes monitoring tool?
It is observability, which begins where traditional monitoring ends. Kubernetes monitoring tells you a metric crossed a threshold: memory climbed, a pod restarted. Moonin starts from that symptom and answers the question that costs time: what changed in the cluster before it happened, which services are affected and what to do next. It does not replace Prometheus for collecting metrics and does not aim to be your alerting system: it uses the cluster’s own deployment history to explain the cause instead of reporting the symptom.
How do I know what changed in my Kubernetes cluster?
Moonin keeps a revision history per service and environment, where every deployment is recorded with its image, its commit and its duration. When something breaks, comparing the last working revision against the current one shows the change without reconstructing it from logs or from kubectl history.
Do I need to instrument my service code?
No. Moonin collects distributed traces and service-to-service traffic with eBPF, in the node kernel, so there are no libraries to add, no SDK, no sidecars and no deployment pipeline changes. You also do not need to install and maintain a separate APM.
Can Moonin modify my cluster?
Almost never. The Discovery Agent and the Node Agent query the Kubernetes API read-only, and the 16 MCP tools it exposes are read-only, with the database session opened in read mode. The single exception is the Scaling Rules Agent, which can adjust HPA and replicas: it is optional, ships disabled, and is enabled by an explicit condition. Moonin does not perform automatic remediation.
Does it work with EKS, GKE and AKS?
Yes. Moonin works against the Kubernetes API, so it behaves the same on Amazon EKS, Google GKE, Azure AKS, self-managed distributions and on-premise clusters. What differs between providers is the node detail each one exposes, not the installation or the data model.
Which DORA metrics does Moonin measure?
It measures the four DORA metrics from DevOps Research and Assessment (deployment frequency, lead time for changes, change failure rate and time to restore service) and adds a fifth view of problem services, ranking services by their contribution to failures. All of them are computed from the cluster’s own deployment history, with nothing to declare by hand.
How much does Moonin cost?
Pricing rests on a single dimension, the Compute Unit, derived from your clusters’ declared capacity rather than from traffic or user count. One CU equals 1 vCPU or 2 GB of RAM, and the heavier dimension is billed, never the sum of both. Pricing is tiered incrementally: USD 20 per CU per month up to 100 units, USD 18 from 101 to 250, USD 15 from 251 to 500, and negotiated above 500. It includes 10 GB of storage per CU and Standard Support; Premium adds 10% and Enterprise 20% on top of the unit cost.
How is Moonin installed?
With a single Helm chart, moonin-agent, applied once per cluster. No sidecars, no deployment pipeline changes, no per-service agent. After installation the inventory of resources, deployments and events appears on its own, continuously.
How much can Moonin reduce a cloud bill?
Arguz measured between 20% and 40% cloud bill reduction across its own projects. The reduction comes from seeing declared capacity nobody uses: extra replicas, oversized requests, and resources left running without an owner.
Does Moonin replace Datadog, New Relic or Grafana?
It solves a narrower problem and costs less because of it: answering what changed in a Kubernetes cluster and what broke as a result. It is not a general-purpose APM or a dashboard engine. It fits when observability spend grew faster than actual usage and what was missing was deployment context, not more time series.