A calm wellbeing self-check
Self-Check
A short, illustrative self-check on everyday rest and balance, built to show how I design careful, accessible, health-adjacent software. It is not a diagnosis and not medical advice. The questions here are examples, not a real clinical tool.
Ten quick questions, one at a time, about the past two weeks. There are no right answers, and nothing you choose is stored. Your result is scored on the server and explained in plain language.
Takes about two minutes. You can go back at any time.
Over the past two weeks, how often did this feel true?
Illustrative questions only. Not a diagnosis and not medical advice.
Your result
Where the signal sits
A breakdown of the same answers across three everyday areas, computed on the server.
lightbulbGentle reflections
footprintPossible next steps
If you need support
This demo cannot help in an emergency. If you are in crisis or worried about your safety, please contact your local emergency services or a trusted crisis line in your area straight away. If everyday strain is weighing on you, speaking with a qualified professional is a good next step.
This is not a diagnosis. If you have concerns, please speak with a qualified professional.
For context
Where your band would sit against an illustrative, made-up spread of results. The numbers below are invented for the demo, not collected from anyone.
How this is built
Server-side scoring
The browser only collects your answers. The server applies the reverse-scoring, theme grouping, and thresholds and decides the band. The client is never trusted to score, and does not even have the rules to.
Accessible by design
Real radio-group semantics, full keyboard support, visible focus states, sufficient contrast, and a reduced-motion mode. One calm question at a time.
Private by default
Nothing you answer is stored. Responses live in memory for the moment it takes to score them, then they are gone. The API key stays on the server.
AI writes, never judges
A language model turns the server-decided band into a plain-language summary against a strict schema. It is told to stay supportive and never diagnose. With no key, a pre-written explainer per band is shown instead.
Honest about what is real: the scoring engine, accessibility, privacy, rate-limited live AI, and the canned fallback are genuine working software. The questions, thresholds, bands, and the context chart are fabricated for this portfolio demo.