Design and operate the Heady Wellness Mirror for personal health tracking, emotional wellness monitoring, and proactive wellbeing coaching. Use when building mood and energy tracking systems, designing stress detection and intervention workflows, creating wellness goal frameworks, implementing longitudinal health pattern analysis, planning privacy-first health data architectures, or designing CBT-aligned coaching interactions. Integrates with headybuddy-core for wellness coaching delivery, HeadyMemory for longitudinal health profiles, heady-observer for wellness alerts, heady-sentinel for health data privacy, heady-vinci for pattern analysis, and heady-patterns for trend detection.
Use this skill when you need to design, build, or operate the Wellness Mirror — Heady's privacy-first personal wellness system that tracks emotional health, energy patterns, and wellbeing signals to provide proactive coaching and self-awareness tools.
The Wellness Mirror operates across Heady's wellness infrastructure:
latent-core-dev, pgvector) — Stores longitudinal wellness profiles, mood histories, and intervention outcomes as 3D vector embeddings (encrypted, user-owned)wellness_mirror:
data_types:
mood:
capture: user-reported (scale 1-5 + optional tags) or inferred (voice tone, interaction patterns)
frequency: configurable (1-3x daily recommended)
dimensions: [valence, arousal, dominance] # PAD emotional model
tags: [anxious, calm, energized, tired, frustrated, grateful, focused, scattered]
energy:
capture: user-reported (scale 1-5) + activity inference
dimensions: [physical, mental, emotional, social]
correlation: tracked against sleep, exercise, work patterns
stress:
capture: composite signal from mood, energy, interaction patterns
indicators: [response_latency_change, vocabulary_shift, session_length_change, error_rate]
levels: [low, moderate, elevated, high, crisis]
sleep:
capture: user-reported (duration, quality) or device-synced
impact: correlated with next-day mood and energy
activity:
capture: user-reported or calendar-inferred
categories: [work, exercise, social, creative, rest, learning]
privacy_model:
principle: user owns all wellness data; platform has zero-knowledge access
encryption: AES-256 at rest, user-held decryption keys
consent: explicit opt-in per data type, revocable at any time
retention: user-configurable (default 1 year, max 5 years, delete on request)
access_control: heady-sentinel enforces strictest-tier protection
sharing: never shared with employers, insurers, or third parties
anonymization: population-level insights use differential privacy (epsilon ≤ 1.0)
audit: every access logged in heady-traces (compliance-grade)
classification: health_data tier in heady-sentinel (highest protection level)
check_in:
delivery:
primary: headybuddy-core conversational check-in
voice: Voice Vessel for hands-free mood capture
quick: HeadyWeb/heady-mobile one-tap mood widget
passive: optional inference from interaction patterns (with explicit consent)
flow:
morning:
1. "How are you feeling this morning?" (mood capture)
2. "How's your energy?" (energy capture)
3. Optional: sleep quality reflection
4. Set intention for the day
5. Store in HeadyMemory wellness namespace
midday:
1. Brief mood check (one-tap or voice)
2. Energy level update
3. Stress signal check
4. Optional: guided breathing if stress elevated
evening:
1. Day reflection: "What went well today?"
2. Mood and energy summary
3. Gratitude prompt (optional)
4. Tomorrow preparation
adaptive_timing:
method: heady-patterns learns optimal check-in times per user
adjustment: shift timing based on calendar, time zone, response patterns
frequency: reduce if user shows check-in fatigue; increase if patterns concerning
nudge: gentle reminder if check-in missed, never punitive
pattern_analysis:
engines:
heady_vinci:
- mood_correlation: identify what activities, people, times correlate with mood states
- stress_prediction: predict high-stress periods from calendar + historical patterns
- intervention_effectiveness: measure which coaching techniques work best for this user
- risk_detection: identify concerning multi-day patterns before crisis
heady_patterns:
- seasonal_trends: detect mood variations across seasons (SAD patterns)
- weekly_cycles: identify day-of-week patterns (Monday dips, weekend recovery)
- trigger_identification: discover specific events that reliably affect mood
- recovery_trajectories: model how quickly user bounces back from low periods
longitudinal:
storage: HeadyMemory time-series wellness embeddings
granularity: daily aggregates + event-level detail
comparison: week-over-week, month-over-month trend analysis
milestones: track improvement over extended periods (3-month, 6-month, 1-year views)
alerts:
engine: heady-observer
triggers:
sustained_low_mood: 3+ days below user's baseline → gentle check-in + coaching offer
energy_crash: sudden drop in all energy dimensions → suggest rest, check workload
stress_escalation: progressive increase over 5+ days → proactive intervention
disengagement: check-in abandonment + reduced platform usage → outreach
crisis_indicators: language patterns suggesting self-harm → immediate safety resources
response:
non_crisis: coaching adjustment, resource suggestions, community connection
crisis: display crisis hotline numbers, suggest professional help, warm handoff guidance
escalation: never automated; always provide human resources, never diagnose