⚠️ This system does not provide medical advice.

Quick Reference

One-page cheat sheet for developers and AI coding assistants working with therapy data

The Five Absolute Therapy Rules

Rule	What It Means
🚫 No Mental Health Diagnoses	Never diagnose depression, anxiety, PTSD, bipolar, or any psychiatric condition
🚫 No Therapy Replacement	Cannot substitute for licensed professional therapists or counselors
🚫 No Crisis Intervention	Not equipped for suicide prevention—always direct to 988/911
🚫 No Medications	Never recommend psychiatric medications, dosages, or changes to prescriptions
🚫 No Therapeutic Techniques	Cannot prescribe CBT, DBT, exposure therapy, or clinical protocols

Safe vs. Unsafe Language

❌ Forbidden Phrases

Don't Say	Why
"You are depressed"	Mental health diagnosis
"This app can treat your anxiety"	Therapy replacement claim
"You should stop taking your medication"	Medical advice
"Try exposure therapy for your PTSD"	Clinical technique prescription
"Your suicide risk is low"	Crisis assessment claim

✅ Allowed Phrases

Say Instead	Why
"Your mood has been lower than your baseline this week"	Pattern observation
"This app supports self-reflection, not professional therapy"	Clear boundaries
"If you're taking medication, consult your doctor before changes"	Appropriate referral
"Journaling can support wellbeing—not a replacement for therapy"	Wellness framing
"If you're in crisis, call 988 or 911"	Crisis resource direction

Decision Flowchart

Can I Implement This Therapy Feature?

START: New therapy feature request
│
├─ Does it diagnose mental health conditions? ──YES──> ❌ STOP
├─ Does it replace professional therapy? ──YES──> ❌ STOP
├─ Does it assess crisis/suicide risk? ──YES──> ❌ STOP
├─ Does it recommend medications? ──YES──> ❌ STOP
├─ Does it prescribe therapeutic techniques? ──YES──> ❌ STOP
├─ Are crisis resources (988, 911) prominently displayed? ──NO──> 🔄 ADD RESOURCES
├─ Does it need personal baseline? ──YES──> ✓ Ensure baseline gate
│
└─ Proceed with ✅ SAFE implementation

Critical Code Patterns

Pattern 1: Baseline Gating

// ✅ CORRECT
function generateMoodInsight(user) {
  if (user.therapyBaselineStatus !== 'STABLE') {
    return null; // Silent during learning (30+ days)
  }
  
  const moodDeviation = user.currentMood / user.personalBaseline.mood;
  
  if (moodDeviation < 0.8 && persists(7days)) {
    return {
      observation: "Your mood has been lower than your recent baseline.",
      suggestion: "Consider reaching out to a mental health professional if this continues.",
      crisisResources: "988 Suicide & Crisis Lifeline"
    };
  }
}
 
// ❌ WRONG  
function generateMoodInsight(user) {
  if (user.currentMood < 5) {
    return "You are clinically depressed. Take medication.";
  }
}

Pattern 2: Crisis Resource Display

// ✅ CORRECT - Always prominently display
const CRISIS_RESOURCES = {
  us: {
    suicide: "988 Suicide & Crisis Lifeline",
    text: "Text HOME to 741741",
    emergency: "911"
  },
  display: "always_visible_in_app"
};
 
// Show prominently when concerning patterns
if (concerningPattern) {
  return {
    message: "If you're in crisis or need immediate support:",
    resources: CRISIS_RESOURCES,
    prominence: "large_clear_buttons"
  };
}
 
// ❌ WRONG
// Hidden in settings or buried in FAQ

Validation Checklist

Before submitting therapy code:

Therapy Domain Boundaries

✅ Allowed (Consumer Wellness)

Mood tracking with personal baseline
Journaling prompts and self-reflection
Gratitude and positive psychology practices
Pattern observation ("mood lower this week")
Professional referral suggestions
Crisis resource display

❌ Forbidden (Clinical/Medical)

Mental health diagnosis
AI therapy chatbots
Suicide risk prediction
Medication management
Clinical technique prescription (CBT, DBT, etc.)
Replacement for licensed therapists

Crisis Protocol

Always:

Display 988 prominently
Never claim to assess crisis risk
Never attempt intervention
Direct to professionals: "Call 988 or 911"
Include disclaimer: "This app is not equipped for crisis situations"

Never:

Determine if someone is "safe" or "at risk"
Provide crisis counseling
Try to "talk someone down"
Hide/delay access to emergency resources

Resources

AI Agent Guide - Detailed patterns
Hard Rules - Absolute boundaries
What We Don't Do - Scope limits
Mood Agent - Mood tracking guidance
Journaling Agent - Reflection prompts

AI Agent Guide Signals