Plain-English investigation
“Why is checkout crashing?” It reads your logs, events and dependencies like a senior SRE to find the cause.
THE SHIP'S MANIFEST
Everything aboard. Every station crewed.
A plain-English AI crew, a one-click app store, and a watch that never sleeps. Here's everything Helmsman does, so you can self-host production Kubernetes without an ops team.
- 55
- one-click apps
- 24/7
- autonomous watch
- 0
- lines of YAML
- 100%
- your hardware
THE HELMSMAN
An ops engineer you can talk to.
Ask in plain English. The Helmsman investigates your cluster like a senior SRE, then hands you the fix as a button. You stay at the wheel.
One-click fixes
Every fix arrives as a button that shows the exact kubectl command before anything runs.
@-mention any resource
Pull a pod, deployment or service straight into the conversation with autocomplete.
Watch it think
A reasoning pane shows how the Helmsman got there, not just the answer.
Choose your model
Opus 4.8 or Sonnet, with Low or High effort. Pick the right model for the job.
Picks up where you left off
Saved chat history, and any panel can hand off to the chat with full context.
THE NIGHT WATCH
It heals the ship while you sleep.
An agent runs inside your cluster, catching drift and failures around the clock and fixing what it safely can, always within the guardrails you set.
Guardrails you define
Spend caps, a kill-switch and scoped permissions. The agent only ever acts inside the lines.
Catches the usual suspects
Crash-loops, OOM kills, pending pods and degraded rollouts, all caught automatically.
Asks when it's unsure
A “Needs you” queue surfaces every decision the agent won't make on its own.
Backs up before it touches
Resource YAML is snapshotted before any change, so nothing is ever lost.
One-click revert
Every action is logged with before / after YAML and an undo button right beside it.
Full audit trail
An activity log of everything the watch has done, ready to review.
THE LOOKOUT
Tell it what to worry about.
Describe what matters in plain English. Helmsman compiles it to a live watch and pings your phone the moment it trips. No query language required.
Alerts in plain English
“Text me if postgres goes down.” No PromQL, no expressions, nothing to learn.
Straight to your phone
Signal and SMS notifications the second something on your watchlist breaks.
Always on the horizon
Your words compile to a live watch that runs continuously in the background.
Nothing extra to run
No Prometheus, no Alertmanager, no dashboards to babysit. Helmsman keeps watch for you.
THE HARBOR
Self-host anything in one click.
The catalog turns “I want to run X” into a live, TLS'd app on your own hardware, with Heroku ergonomics and none of the cloud bill.
55 apps, six categories
Dev tools, databases, productivity, media, networking and observability, all ready to deploy.
Guided install
Pick namespace, hostname, storage and node fit, with every choice validated before it runs.
TLS, issued & renewed
Every app gets HTTPS automatically. Certificates issue and renew themselves.
Stay current
Update checks spot new versions and flag upgrades the moment they land upstream.
Clean uninstall
The Grand Cleanse finds every resource, confirms by typed name and purges without a trace.
On your hardware
Heroku ergonomics, docked at your cluster. You keep control of your data and your bill.
THE BRIDGE
Every resource, one console.
No flipping between kubectl get and a wall of YAML. Every core Kubernetes resource gets its own panel:
searchable, namespace-scoped, and one click from a fix.
Workloads
Deployments, StatefulSets, DaemonSets, Jobs & CronJobs
Pods
Phase, restarts and live CPU / memory sparklines
Services
Ports, endpoints and type at a glance
Ingresses
Hosts, TLS and backend routing
ConfigMaps
Keys, values and sizes, searchable
Secrets
Masked values, types and sizes
Storage
PVCs, access modes and capacity
Namespaces
Create, delete and scope the whole console
Nodes
Capacity, status, cordon & drain
RBAC
Roles, bindings and service accounts
Certificates
cert-manager status and renewals
Events
A live feed of what the cluster is doing
THE ENGINE ROOM
Drop into the machinery.
When you want to get hands-on, every common operation is a click away. No kubectl incantations to memorize, no YAML to edit by hand.
Port-forward
Tunnel any service or pod to localhost, with local-port conflict detection.
Live logs
Stream pod logs in real time over a WebSocket, with configurable tail.
Scale & restart
Scale, restart, pause and resume rollouts, each from a single button.
Edit env & images
Set variables, swap container images and adjust requests & limits in place.
Node operations
Cordon, uncordon and drain nodes safely, with sensible defaults baked in.
CronJob control
Suspend, resume or trigger an on-demand run straight from the schedule.
THE CHARTS
The view from the crow's nest.
Read the state of the whole cluster at a glance, then send the Helmsman to investigate anything that looks off.
Overview dashboard
Node, pod and deployment health at a glance, with a one-click “investigate this”.
Connectivity map
See pod-to-pod and pod-to-service flows and spot where traffic is unhealthy.
Database radar
Auto-detects Postgres, MySQL and Supabase. Copy a DSN, reveal creds or port-forward.
Right-sizing
Flags over- and under-provisioned workloads and suggests the CPU / memory fix.
SECURE BY DEFAULT
You're always at the wheel.
Helmsman can do a lot, but it never acts behind your back. Approval is explicit, access is yours alone, and everything runs on your own infrastructure.
Nothing runs unconfirmed
Every mutation shows the exact kubectl command in a confirm sheet before it executes.
Password & sessions
Built-in login with httpOnly session cookies. Secure by default, no extra proxy required.
Your keys, your cluster
Self-hosted end to end. Your data and credentials never leave the hardware you control.
Namespace scoping
One shared namespace filter scopes the entire console to exactly what you want to touch.
LEAVE IT ON THE DOCK
All the parts you don't have to carry.
The usual self-hosting toolbelt stays ashore. Helmsman handles the machinery so you can describe what you want in plain English.
no YAML to hand-write no PromQL no Prometheus to run no Helm CLI no kubectl memorized no dedicated ops team
Take the helm.
Self-host like you have an ops team: the crew you couldn't afford, diagnosing, fixing, and standing watch around the clock.