What this prompt does
This prompt casts the AI as a senior frontend engineer building a patient dashboard, and it asks for working component code and a layout spec rather than a mockup description. It specifies an appointments card, current medications with a refill CTA, a vitals summary with trend charts, recent labs with abnormal-value flagging, and a secure message inbox. The [vitals] placeholder decides which trend charts are generated, and [compliance] sets the accessibility and privacy bar the components must meet.
The structure works because health data raises the floor on both accessibility and privacy, and the prompt bakes both in. It requires semantic landmarks, focus order, ARIA on charts, and color-independent abnormal flags, so flags never rely on color alone, which fails real users and audits. The [brand] variable sets accent and tone, the [stack] keeps components aligned to your framework, and a dedicated note marks where PHI must never be cached, logged, or sent to analytics. Pinning the PHI boundary up front matters because once an analytics SDK is wired in it is easy to leak protected data through event payloads or logs without noticing, and unwinding that later is far harder than drawing the line at design time.
When to use it
- Building a patient portal or health dashboard with vitals and labs
- Charting a specific set of
[vitals]like blood pressure, glucose, or weight - You need a
[compliance]bar such as WCAG 2.2 AA and HIPAA-aware baked in - Flagging abnormal lab values without relying on color alone
- Defining PHI boundaries before wiring any analytics SDK
- Generating components in a known
[stack]with a clear layout spec
Example output
You get component files, a layout spec, and an accessibility/PHI note. The dashboard lays out appointments, medications with refill CTA, a vitals summary with trend charts for your listed vitals (each with empty and loading states), recent labs with abnormal-value flagging, and a secure inbox. The PHI note calls out where protected health information must never be cached, logged, or sent to analytics, and the components carry semantic landmarks and ARIA on charts.
Pro tips
- Set
[vitals]to exactly what you track; one trend chart is generated per vital, so the list drives the charts - Use
[compliance]to state your real bar; WCAG 2.2 AA plus HIPAA-aware changes the markup the model produces - Decide your no-log, no-cache PHI boundaries from the PHI note before you add any analytics SDK
- Match
[stack]to your framework so the component files are directly usable - Ask it to confirm abnormal flags use an icon or label, not color alone, since color-only flags fail audits
- Use
[brand]to keep the clinical tone reassuring rather than alarming across the dashboard
