Free path pitch - "look before you install"

A repository document, rendered for the site. View source markdown.

What this command does. cub installer is a released, open-source plugin for the cub CLI. cub installer setup pulls a catalog package and writes its Kubernetes files locally. It does not apply those files to a cluster; use kubectl, Argo CD, or Flux for delivery. The generated scripts stop before doing any work when the plugin or kustomize is missing.

New to cub? Install the cub CLI first. You can pull and render public catalog packages without an account. Commands that save or change ConfigHub data require you to sign in.

Generated at: 2026-07-30T12:38:02.000Z UTC · source: committed helm-expt evidence for this rendered repository document.

UNOFFICIAL/EXPERIMENTAL.

The thesis

Lead the free path with a pre-flight check - pure upside, no migration - not with parity. Parity drops to a quiet line of assurance underneath. This answers the critique that "parity sounds like a solution to a problem this approach itself creates": if the only reason you need parity is that we introduced cub, leading with it is damage control. The pre-flight check is a real pareto gain instead - benefit with no offsetting loss, because you change nothing and switch nothing.

Safety and correctness are the spine through all three tiers: free see it → server fix it once and propagate → paid govern it.

Ship now - the shipped subset

Every claim below is backed by something we already do. This is the copy that can go on the site today.


Look before you install

Helm installs a chart in one step and tells you what it did afterward. Render it with cub first - free, no account, without changing how you install - and read what you're about to run.

See every object. The exact Kubernetes resources the chart will create, as plain files you can read before anything reaches the cluster.

See the secret it bakes in. Many charts generate a password and ship it inside the render. Spot it, and swap in your own, before it lands in your cluster or your registry.

See what your cluster needs first. The CRDs, secrets, and targets that must already be there - named up front, instead of a half-applied release to debug.

Bringing AI? When AI writes your values, render what it produced before it's live - the objects on screen, not a wall of plausible YAML you take on faith.

You switch nothing. Keep Helm. Keep Argo or Flux. Keep your AI. You look first - and there's nothing to undo, because you changed nothing.

And it's provably the same objects Helm would install. We're not changing your install. We're showing you what's in it.


What backs each line (so it is not an overclaim):

North star - the full pitch, once the builds land

Same copy, with three more lines that need building first (marked ⏳):

The three builds that make the full pitch true

  1. Render-time misconfig scan - open ports, privileged containers, missing auth, weak defaults (kubescape / checkov-style), run on the rendered objects. Turns "insecure defaults" from the baked-secret slice into the real security hook.
  2. Upgrade pre-flight / CRD-skew check - detect that a chart bump drops an in-use CRD version, then offer a version both releases support. (A concrete feature proposed by a colleague; chkk-shaped, scoped to a chart.)
  3. Grounded AI explainer - "why won't this render?" answered from the actual render plus the catalog's per-chart knowledge. The grounding data exists; the explainer is the build.

These also extend the safety spine into server and paid: gate AI/misconfig changes across a fleet (server), governed agentic remediation with compliance and SLA (paid).

Placement and generator notes (for Codex)