Skip to main content
To get value from RubixKube, connect at least one environment. The Observer lives near your workload, discovers topology, and streams signals to RubixKube Cloud for correlation. A single workspace can hold any mix: one EKS cluster, one AWS account, one GCP project, and a handful of VMs, all correlated in the same knowledge graph.
Connect your infrastructure picker with tiles for Kubernetes, Amazon AWS, Google Cloud, Bare Metal or VMs available today, and Microsoft Azure, Oracle Cloud, DigitalOcean, VMware, and OpenStack marked coming soon.

​
Pick your path

Kubernetes

EKS, GKE, AKS, KIND, or any conformant cluster on v1.24 or later. Install via kubectl apply.

AWS

Account-level observation across EC2, RDS, Lambda, S3, ELB, CloudTrail, and CloudWatch.

GCP

Project-level observation across GCE, GKE, Cloud SQL, Cloud Run, Cloud Storage, and Cloud Functions.

Linux VMs

Any modern Linux host. CPU, memory, disk, network, and per-process signals.
Running on Azure? AKS clusters are supported today via the Kubernetes path. Subscription-level Azure observation (App Service, Azure VMs, managed data) is on the roadmap. Email us to join the early-access list.

​
The install, in three moves

Every environment follows the same shape. The per-environment pages above cover the specifics.
1

Copy the personal command from the console

Open Environments in console.rubixkube.ai, create a new environment, and copy the install command. Your API key (starting with rk_) is already embedded, so you can paste and run.The shape looks like this:
kubectl apply -f "https://api.rubixkube.ai/install/observer.yaml?apiKey=rk_YOUR_API_KEY"
Always copy the command from the console. The embedded API key scopes the Observer to your workspace.
2

Run it against the target environment

The Kubernetes manifest applies in one shot. The shell installer is interactive and asks where to deploy:
What would you like to do?
  1) Install Observer
  2) Uninstall Observer
> 1

Where do you want to deploy the observer?
  1) On this machine
  2) Create a new AWS EC2 instance
  3) Create a new GCP Compute Engine instance
> 1
Pick option 1 to install on the current host, option 2 to spin up a fresh EC2 instance, or option 3 for a fresh GCE instance. Follow the region and project prompts from there.
3

Verify it is running

kubectl get pods -n rubixkube-system
Wait for STATUS to show Running.
In the console, the environment card switches from Connecting to Healthy within a minute or two and the Infrastructure view populates with a live topology.

​
What the Observer does

The Observer is deliberately light: it collects signals, maintains a local cache, and streams to RubixKube Cloud. It never sends raw payloads outside of what you configure.
  • Discover. Walks your environment through read-only APIs and builds a live topology.
  • Stream. Sends structured events, metrics, and state snapshots to RubixKube Cloud over HTTPS and NATS.
  • Stay out of the way. Read-only by default. No mutations without explicit approval, enforced by the Guardian agent.

​
Requirements at a glance

EnvironmentInstall surfaceOutbound networkTypical footprint
Kuberneteskubectl apply manifestapi.rubixkube.ai:443, nats.rubixkube.ai:443~255Mi RAM, under 10 mCPU
AWSShell installer on a VM, or new EC2Above, plus AWS APIs~200Mi RAM
GCPShell installer on a VM, or new GCEAbove, plus GCP APIs~200Mi RAM
Linux VMShell installer on the hostapi.rubixkube.ai:443, nats.rubixkube.ai:443~150Mi RAM

​
After you connect

Monitor infrastructure health

Set up the dashboard, insights, and notifications for your team.

Understand Environments

Manage multiple environments, review health, and reorganise workspaces.

Set up integrations

Connect Slack, Linear, PagerDuty, and more.

Safety and Guardrails

Understand what RubixKube will and will not do on its own.

​
Common questions

Yes. Each connection is independent, so you can run them in parallel. Free tier allows one environment, Business allows up to three, Enterprise is unlimited.
Read-only by default, scoped to the platform you are monitoring. Anything that could change state needs explicit approval and is bounded by Guardian policies. See Safety and Guardrails.
Structured events, metrics, and state snapshots stream to RubixKube Cloud for correlation. Raw payloads stay in your environment unless you explicitly enable them.
The Observer only needs outbound HTTPS to api.rubixkube.ai:443 and nats.rubixkube.ai:443. If that is not possible, reach out and we can talk through a deployment mode that fits your controls.
Delete it from Environments in the console, then run the appropriate uninstall. For Kubernetes: kubectl delete -f "https://api.rubixkube.ai/install/observer.yaml?apiKey=rk_YOUR_API_KEY". For shell installs: rerun the installer and pick Uninstall Observer.