About

Built from a thousand context switches.

Why I made Klystiq, and who it's for.

It's two in the morning, and checkout is down in Europe. The same Deployment is perfectly healthy in the US. Same image, same chart, same everything, supposedly.

I have seven terminal tabs open. One is tailing logs in prod-eu-west. Another is tailing logs in prod-us-east, and I keep forgetting which is which. A third is running kubectx so often the history is just the word "prod". I'm piping two kubectl get -o yaml outputs into files so I can diff them, and the diff is mostly noise: resource versions, timestamps, managed fields. The customer's team is on a call, watching me scroll.

We found it eventually: a ConfigMap someone had changed by hand in one region, months earlier. Three lines. It took us an hour and forty minutes to see three lines.

That night wasn't unusual. That was the problem.

Hundreds of teams, one pattern

I've spent years helping companies build and run their platforms on Kubernetes: hundreds of customers, from two-person startups with one cluster they were slightly afraid of, to enterprises with dozens of clusters across clouds, regions and teams. Every one of them was different, and every one of them ended up the same way: Kubernetes is multi-cluster by default. Production per region. Staging that's supposed to match production. Dev clusters, team clusters, the cluster nobody admits to owning.

The tools, though, mostly think about one cluster at a time. You look at one, then switch, then look at the other one, then try to hold both in your head.

Free tools, commercial tools, and the gap between them

I tried everything, usually on someone else's laptop, under pressure. The free tools are genuinely impressive, and I still use kubectl every day. But a terminal is one cluster and one question at a time, and the web dashboards wanted to be installed into the cluster, with agents, service accounts and a security review that took longer than the project.

The commercial tools were better looking and often better thought out. Then I'd hit the wall: the feature I needed was a tier up. Or it wanted me to create an account and sign in, to a tool whose whole job is to read the kubeconfig sitting on my disk. Or it phoned home. Or it was a browser in a trench coat that took a gigabyte of memory to show me a table. Some of my customers, in finance and healthcare and government, couldn't use them at all: no cloud accounts allowed, no telemetry allowed, sometimes no internet.

None of them fitted the way I actually work, and none of them fitted the way my customers' engineers worked either. I kept writing little scripts to fill the gaps, and kept handing them to customers, and kept watching them write their own.

So I built the tool I kept wishing for

Klystiq started as a weekend experiment: a native app, in Rust, that would open every context in my kubeconfig and show me all the Pods in one table. That was it. The first time I saw prod-us-east and prod-eu-west side by side, sorted and filterable in one list, I found a misconfigured readiness probe I'd walked past for a week. I didn't go back.

Everything since has come from the same place: a question I or a customer kept asking. Why does staging work and production doesn't? became Compare namespaces. Why did the cluster stop admitting Pods? (a webhook's CA certificate had quietly expired) became the Certificates page. What changed in the last hour? became Timeline. Will this upgrade break anything? became the Upgrade check. And every dialog shows the kubectl command it's equivalent to, because half my job has always been teaching, and the best tool makes you better at the underlying one, not dependent on it.

It runs on your computer and nowhere else. It reads the kubeconfig you already have and never changes it. It installs nothing in your clusters, asks you to sign in to nothing, and sends nothing anywhere unless you ask. It's careful with production, because I've been the person on the 2 a.m. call, and I've seen what one wrong context can do.

Fair to the people who use it

Klystiq is free for personal use, with every feature. If you're learning Kubernetes, running a home lab or maintaining an open-source project, you'll never see a paywall. When you use it at work, a license keeps it going, and the price is meant to be an easy yes for the people who sign off on it. Every tier gets every feature, because I've been on the wrong side of "that's in Enterprise" too many times.

Klystiq is made by Mark Richman LLC, which right now means me. I read every email, and a lot of the app exists because someone wrote to say "I wish it did this". If that's you, I'd love to hear from you.

Mark RichmanCreator of Klystiq

What I care about

The principles behind Klystiq.

Every cluster at once

Real teams run many clusters. The tool should start from that, not bolt it on.

Local by design

Your clusters, your credentials, your computer. No account, no agents, no one watching.

Careful by default

Dry runs, diffs and read-only contexts, because the scariest command is the one you didn't mean to run.

Teach, don't hide

Every action shows the kubectl it's equivalent to. A good tool should make you better at the real thing.

Keyboard first

If your hands have to leave the keys during an incident, the tool is in the way.

Every feature, every tier

Free for personal use with nothing held back. Companies pay for the right to use it at work, not to unlock it.

Say hello.

Questions, ideas, or a story about your own 2 a.m. call: write to support@klystiq.com.