Think inside your AI world.

Kubernetes through Unl

Kubernetes shows the pod restarting and the logs behind it. Whether that's a rollout doing its job or an incident is your rule - and 'it's not Running' isn't the same as 'it's broken'.

Kubernetes through Unl cross-references pod status and logs against the rollout tolerance the platform engineer has ratified, so 'is this crash loop expected?' returns verdict-shaped, naming what hasn't settled.

What Kubernetes holds

What the cluster actually hands back:

  • Pod status across every namespace in the cluster
  • Logs for a named pod
  • Any resource by kind - deployments, services, config, events
  • Cluster context when more than one cluster is in scope

What the naked read gives you

The Kubernetes MCP server reads the cluster directly - pod status across every namespace, logs for any pod by name, any resource by kind with a real Get or List call. It's a native implementation talking to the API server, not a kubectl wrapper pretending. What it returns is state, not verdict: a pod that isn't Running could be mid-rollout, waiting on a dependency, or genuinely stuck, and the API surface doesn't distinguish those from each other. The pod restarted three times. Whether that's expected is a judgement the cluster was never asked to hold.

The frame judges the data it is given; it does not verify the source’s accuracy.

What changes when Kubernetes is read measured

The platform engineer ratified in Unl that a pod's crash loop only counts as an incident when the restart count exceeds three within ten minutes and the deployment isn't mid-rollout - because rolling restarts were paging the team for their own release process.

'Is this crash loop expected?' returns, measured: no - the deployment finished rolling out six minutes ago, and this pod is still restarting outside that window. Kubernetes supplied the pod status and restart count; your ratified rule supplied the rollout window that separates expected churn from an incident.

And back again

The read loops back: a crash-loop verdict - expected rollout churn or genuine incident - gets appended to Unl with the pod, the restart count and the rollout window it was checked against. The next time this deployment misbehaves, the platform engineer isn't reasoning from zero; the prior verdict and its window are already there to confirm, tighten, or challenge.

The answer comes back measured against what you already decided, and why.

The lane is live and open to this tool today: one box, paste anything. If it speaks MCP, Unl can reach it. Readings arrive unprompted, the data beside the criterion; Unl is a courier, not a warehouse, and keeps only your keys and the frame.

Questions people ask

Can AI tell me if a crash loop is actually a problem?

Not on read data alone. Kubernetes through Unl reads the pod status, restart count and logs; whether that counts as an incident depends on the rollout window you ratified in Unl. The same restart count returns a different verdict once the deployment has settled.

How do I connect Kubernetes to Claude?

The containers project maintains the Kubernetes MCP server at github.com/containers/kubernetes-mcp-server. Point it at a kubeconfig and it talks to the API server directly; Unl holds the rollout tolerance the pod status is measured against.

Does Unl restart or delete anything in the cluster?

No. What comes back through the Kubernetes MCP server is pod status, resource detail and logs - a snapshot of the cluster, not a lever on it. Unl doesn't send restart, scale or delete commands.

What this is

Think inside your AI world — you stay in command

Unlimitless (Unl to friends) holds what you've settled, reads what your tools are showing, and catches what's changed out in the world — and hands your AI whatever bears on the work, the moment it's needed, without you asking. The right thing, in front of the model, unprompted, with you in command of the call. So you keep moving toward what you set out to build, on top of everything you've already decided.

It plugs into Claude, Claude Code, ChatGPT and Cursor as an MCP connector. Quick to connect, in a couple of steps.

Unlimitless is open now to invited Alpha. Apply for the Beta waitlist to come in ahead of the full launch:

Alpha is invite-only · free at launch.