Paste this prompt into Claude Design when you want to audit ND accessibility.
Cytognosis context: nd-design-guide.md, 11_accessibility.md, companion-profile.md
Brief
Build a comprehensive neurodiversity accessibility testing protocol for Cytognosis surfaces. This covers automated checks, manual testing with ND participants, heuristic evaluation, and WCAG 2.2 criteria mapping.
1. Automated Checks (CI-Blocking)
Run these checks on every PR that touches CSS, HTML, or JSX:
| Check | Tool | Threshold |
|-------|------|-----------|
| Text contrast ratio | axe-core | 7:1 body, 4.5:1 heading (WCAG AAA) |
| Touch target size | axe-core | 48px Companion, 56px Crisis |
| Font size minimum | stylelint | 17px body Companion, 18px Crisis |
| Line height minimum | stylelint | 1.5 general, 1.7 Companion |
| No pure black/white | stylelint | Reject #000, #fff, #000000, #ffffff |
| No italic emphasis | stylelint | em { font-style: normal } enforced |
| No justified text | stylelint | Reject text-align: justify |
| Animation duration | custom | 0ms when [data-motion] absent or disabled |
| No autoplay | custom | Reject autoplay attribute on media elements |
| Focus visible | axe-core | All interactive elements have visible focus |
2. Manual ND Testing Protocol
2.1 Participant Criteria
Minimum 5 participants with self-identified neurodivergent conditions
Mix: ADHD (2+), dyslexia (1+), autism spectrum (1+), anxiety/depression (1+)
Ages 18-65, varying tech literacy
Compensate fairly ($75/hour minimum)
2.2 Session Design
Maximum 45 minutes (cognitive fatigue boundary)
Breaks every 15 minutes (offer, do not require)
Quiet, low-stimulation environment
Screen recording with consent; no live observers behind glass
Provide task list in writing before session starts
Allow participants to skip any task without explanation
2.3 Task Categories
A. First Impressions (5 minutes)
Show the main screen for 10 seconds, then ask: "What do you notice first?"
"Does anything feel overwhelming or calming?"
"Can you find the main action?"
B. Core Flow (15 minutes)
Complete the primary task (e.g., log mood, review biomarker)
Note: time to first action, errors, hesitations, re-reads
Ask: "Was anything confusing?" after each step
C. Customization (10 minutes)
Find and use the font toggle
Find and use the density control
Find and use the motion toggle
Ask: "Which settings feel most comfortable?"
D. Notification Response (5 minutes)
Trigger a test notification
Measure: time to notice, time to act, emotional response
Ask: "How did that feel? Too much? Too little?"
E. Recovery (5 minutes)
Present an error state
Can the participant recover without help?
Is the next action clear?
3. Heuristic Evaluation Checklist
3.1 Cognitive Load
[ ] Can the user complete the primary task without reading instructions?
[ ] Is there only one primary CTA visible at a time?
[ ] Are paragraphs 4 sentences or fewer?
[ ] Is content width 65ch or less?
[ ] Are form fields shown one at a time (wizard pattern)?
[ ] Is progress visible (step indicators, progress bars)?
3.2 Distraction Audit
[ ] Is there any moving element not initiated by the user?
[ ] Are there any auto-playing media elements?
[ ] Is the notification count 1 or fewer at a time?
[ ] Are decorative elements minimal and non-distracting?
[ ] Is the color palette muted (300-shade scale)?
3.3 Navigation Predictability
[ ] Is the nav structure identical on every page?
[ ] Are buttons in consistent positions across screens?
[ ] Does every action produce immediate visual feedback?
[ ] Are state indicators always visible (where am I, what is happening)?
3.4 Typography Comfort
[ ] Can the user read for 10 minutes without eye strain?
[ ] Is bold the only emphasis method (no italic, no underline)?
[ ] Is text left-aligned (never justified)?
[ ] Is the default font reading-fluency optimized (Lexend)?
3.5 Error Recovery
[ ] Does every error state offer a clear next action?
[ ] Are error messages non-punitive and specific?
[ ] Can the user undo any destructive action?
[ ] Is data auto-saved so nothing is lost?
4. WCAG 2.2 Criteria Mapping
| Criterion | Description | Target | Profile |
|-----------|------------|--------|---------|
| 1.4.3 | Contrast (minimum) | 4.5:1 text | All |
| 1.4.6 | Contrast (enhanced) | 7:1 body text | Companion, Crisis |
| 1.4.12 | Text Spacing adjustable | User density control | Companion |
| 2.2.2 | Pause, Stop, Hide | All motion user-controlled | Companion |
| 2.4.6 | Headings and Labels | Descriptive headings | All |
| 2.5.5 | Target Size (enhanced) | 44px+ (48px Companion, 56px Crisis) | Companion, Crisis |
| 3.2.3 | Consistent Navigation | Nav identical across pages | All |
| 3.3.2 | Labels or Instructions | Visible label on every field | All |
| 3.3.7 | Redundant Entry | No re-entering previously provided info | Companion |
| 3.3.8 | Accessible Authentication | No cognitive function tests | All |
5. Reporting Template
After testing, produce a report with:
1. Summary: pass/fail counts by category
2. Critical findings: any issue that blocks task completion
3. Comfort findings: issues that cause discomfort but not failure
4. Participant quotes: direct feedback (anonymized)
5. Recommendations: prioritized fix list with effort estimates
6. Before/after screenshots: for each recommended fix
7. Retest plan: timeline for verifying fixes
6. Frequency
| Trigger | Test Type |
|---------|-----------|
| Every PR (CSS/HTML/JSX) | Automated checks (CI) |
| Monthly | Heuristic evaluation (internal) |
| Quarterly | Manual ND testing (external participants) |
| Major release | Full protocol (automated + heuristic + manual) |
| New profile or component | Manual ND testing for that surface |