UNOFFICIAL/EXPERIMENTAL. Persona UX testing checks whether a human can understand and use the public site. It complements proof lanes. It does not replace render, live, or ConfigHub evidence.
Issue: https://github.com/confighub/helm-expt/issues/1018
Scope
By default, this audit covers the free/public website and documentation surface: home, Get Started, Helm Catalog, chart pages, Variants, Apps, Ops, Docs, FAQ, and generated public data linked from those pages. Commercial and managed-edition flows are out of scope unless a run explicitly asks for them.
When To Run
Run this after major public-site, user-doc, homepage, catalog, FAQ, or Get Started changes.
Depth Requirement
Do not stop at the homepage or top navigation. Each persona run must include a depth pass through the pages the persona would naturally reach within one or two clicks of home:
| Entry point | One-click pages | Two-click pages to sample |
|---|---|---|
| Home | Get Started, Helm Catalog, Variants, Apps, Ops, Docs, FAQ | Redis chart page, kube-prometheus-stack chart page, How It Works, Known Gaps, AI-assisted changes, GitOps adopter guide, upgrade/rollback guide |
| Helm Catalog | chart index, chart page, matrix link | chart recipe/catalog link, per-chart caveats, lifecycle/action route, source chart URL |
| Apps | existing-app section, app examples | adopting existing apps, ConfigHub data model, custom app or stack guide |
| Ops | observe, rollback, delivery, scan/gate cards | verify-it-yourself, day-2 upgrade/rollback, why-synced-is-not-working, cub-scout diff design |
| FAQ | skeptical questions | the guide or data page linked from each answer |
For every deeper page sampled, record whether the user can still answer:
What am I trying to do?
What should I type?
What should I expect to see?
Do I need a cluster, ConfigHub account, GitOps controller, or only local files?
Is this a proved path, a watch path, a blocked path, or planned work?
Where do I go next?If a page answers only by linking to raw data, treat that as a possible bounce unless the persona is the skeptical reviewer.
Personas
| Persona | What they want |
|---|---|
| Novice Kubernetes user | Learn what this is without knowing ConfigHub. |
| Low-skill Helm user | Copy commands and see expected output. |
| Helm expert | Understand where values, hooks, CRDs, and upgrades go. |
| GitOps platform engineer | Keep Argo or Flux and know what changes. |
| SRE or on-call engineer | Triage what broke and what evidence exists. |
| Chart maintainer | See whether their chart can be represented honestly. |
| AI-curious product lead | Understand what AI can safely help with. |
| Skeptical reviewer | Find limits, watch rows, refusals, and proof scope. |
Seed Questions
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.
What is ConfigHub?
Do I need an account?
Do I need a cluster?
Can I use my existing Argo or Flux setup?
Can I load my existing app or platform?
How do I know cub installer preserved Helm behavior?
Where did my Helm hook go?
What output should I expect after each step?
What is safe for AI to change?
What do I do when a chart breaks?Run Format
For each persona, record:
persona
task or fear
path tested
deeper pages tested
where the site answered well
where the persona would bounce
improvement candidate
voice owner: marketing voice or technical voiceVoice Split
Use marketing voice for intros, landing pages, user motivation, and marketing content. It should be plain, direct, optimistic, and human.
Use technical voice for technical content. It should define scope, prerequisites, current status, evidence, limits, and next actions precisely.
Output
The output is a ranked website plan, not a new proof surface. Keep the proof links underneath the user journey.
Recorded runs: