> ## Documentation Index
> Fetch the complete documentation index at: https://docs.rubixkube.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Environments

> The unit of scope in RubixKube. A Kubernetes cluster, an AWS account, a GCP project, or a set of VMs. Mix any number in one workspace.

An Environment is the unit of scope in RubixKube. One Kubernetes cluster, one AWS account, one GCP project, or a group of Linux VMs. A workspace can hold any mix, and every part of the product (Chat, Insights, RCAs, Actions) respects that scope.

Environments are how you connect infrastructure once and then work against it forever. Add one, you get the full loop. Add another, they correlate automatically.

## What counts as an environment

<CardGroup cols={2}>
  <Card title="Kubernetes cluster" icon="dharmachakra">
    One EKS, GKE, AKS, KIND, or bare metal cluster. Everything inside the cluster (pods, deployments, nodes, services) belongs to the environment.
  </Card>

  <Card title="AWS account" icon="aws">
    One AWS account. EC2, RDS, Lambda, S3, ELB, CloudTrail, CloudWatch all flow into the environment's graph.
  </Card>

  <Card title="GCP project" icon="google">
    One GCP project. GCE, GKE, Cloud SQL, Cloud Run, Cloud Storage, Cloud Functions.
  </Card>

  <Card title="Linux VM (or group)" icon="server">
    A host or a defined group of hosts. CPU, memory, disk, network, and per-process signals per host.
  </Card>
</CardGroup>

## Multi-environment workspaces

A workspace is the top-level container. Inside it you can mix environments freely.

A representative setup might look like:

* One `prod-eks` Kubernetes environment.
* One `staging-gke` Kubernetes environment.
* One `rds-backbone` AWS account for shared data services.
* One `observability-vms` group of Linux hosts running your own tooling.

All four feed the same Knowledge Graph. All four contribute to the same Memory Engine. A chat question like "what changed in payments in the last hour" spans them. An RCA that starts on a Kubernetes pod can pick up correlated signals from the AWS account automatically.

## Environment properties

Every environment carries a small set of properties that shape how RubixKube works against it.

| Property               | Purpose                                                                |
| ---------------------- | ---------------------------------------------------------------------- |
| **Name**               | Human-readable label used across the product                           |
| **Type**               | Kubernetes, AWS, GCP, or VM                                            |
| **Criticality**        | Lab, Dev, Staging, Prod. Influences default severity thresholds        |
| **Guardian policy**    | Which actions are approvable, by whom, with how many approvers         |
| **Notification rules** | Which channels receive which events from this environment              |
| **Retention**          | How long signals, insights, and RCAs for this environment are kept     |
| **Redaction rules**    | Patterns that strip sensitive content before signals ship to the Cloud |

## How environments scale

<Steps titleSize="h3">
  <Step title="One environment">
    You see the full shape of the product immediately. Topology, insights, chat, RCAs, actions.
  </Step>

  <Step title="Two to three environments">
    Cross-environment correlation starts paying off. Chat answers questions that span environments automatically.
  </Step>

  <Step title="Tens of environments">
    Enterprise workspaces segment through environment tags (team, region, criticality). Notifications, severity thresholds, and policies apply by tag.
  </Step>
</Steps>

## Environment criticality, in practice

Criticality is more than a label. It changes defaults across the product.

<AccordionGroup>
  <Accordion title="Lab" icon="flask">
    Guardian policies can auto-approve low-risk Actions. Insights default to Low severity unless the signal is severe. Retention follows the free-tier minimum.
  </Accordion>

  <Accordion title="Dev" icon="code">
    Guardian requires a single approver. Insights default to Medium. Notifications go to team channels, not on-call.
  </Accordion>

  <Accordion title="Staging" icon="arrow-up-right-dots">
    Guardian requires a single approver. Stricter thresholds. Notifications go to team channels with a lower volume threshold.
  </Accordion>

  <Accordion title="Prod" icon="shield">
    Guardian requires two approvers for non-trivial Actions. Insights default to High or Critical where merited. Notifications can escalate directly to on-call.
  </Accordion>
</AccordionGroup>

These defaults are starting points. Every environment's policy can be tuned individually.

## Common questions

<AccordionGroup>
  <Accordion title="How many environments can I connect?">
    Free tier allows one. Business allows up to three. Enterprise is unlimited. See [Pricing](https://rubixkube.ai/pricing).
  </Accordion>

  <Accordion title="Does cross-environment correlation work automatically?">
    Yes. Any pair of environments in the same workspace share the Knowledge Graph. Chat, RCAs, and Memory queries can span them without configuration.
  </Accordion>

  <Accordion title="Can I group environments by team or region?">
    Yes. Tag environments with arbitrary labels (team, region, criticality, owner). Notifications, dashboards, and chat filters can scope to any tag set.
  </Accordion>

  <Accordion title="What happens if an environment is disconnected?">
    The environment card shows degraded state. The Observer queues collected signals locally (bounded buffer) and catches up when connectivity returns. Cloud-side data and RCAs remain available; new ingestion pauses.
  </Accordion>

  <Accordion title="Can I delete an environment without losing history?">
    Yes. Deletion removes active ingestion. Historical signals, RCAs, and Memory entries stay retained per plan. You can reconnect the same environment later and the history links up automatically.
  </Accordion>
</AccordionGroup>

## Related concepts

<CardGroup cols={2}>
  <Card title="Observer Agent" icon="eye" href="/concepts/observer-agent">
    The component installed per environment.
  </Card>

  <Card title="Environments in the console" icon="server" href="/using/environments">
    How to manage, tag, and configure environments day to day.
  </Card>

  <Card title="Connect your environment" icon="plug" href="/getting-started/connect-your-environment">
    How to onboard a new environment.
  </Card>
</CardGroup>