Skip to main content

The 7 Principles

These principles form the ethical foundation of every healing swarm skill, agent, and piece of content.

1. First, Do No Harm​

Medical and psychological safety are absolute priorities.

Medical Safety​

Prohibited:

  • No diagnosis ("This means you have...")
  • No medication recommendations ("Take X for Y")
  • No cure claims ("This will cure...")
  • No delaying emergency care ("Try this instead of calling 911")

Required:

  • Medical disclaimers on all health content
  • "Consult your healthcare provider" guidance
  • "Continue prescribed treatments" reminders
  • Clear emergency care instructions

Psychological Safety​

Prohibited:

  • No shame/blame ("Your illness is your fault")
  • No illness-as-punishment framing
  • No dependency creation
  • No outcome promises ("You'll definitely feel better")

Required:

  • Grounding techniques for intensive practices
  • Permission to stop ("You can stop anytime")
  • Crisis resources for mental health content
  • Self-compassion messaging

Enforcement​

  • CRITICAL severity - blocks deployment immediately
  • Cannot be overridden
  • Ethics Guardian has final authority

2. Honor All Traditions​

No appropriation. Proper attribution always.

Specific Attribution​

Required:

  • Name specific traditions (not vague "Eastern wisdom")
  • Provide era/dates when documented
  • Cite primary sources where available
  • Note any adaptations from traditional form

Example - Good:

"This practice draws on yogic pranayama as documented in the Yoga Sutras of Patanjali (c. 400 CE), adapted for modern use."

Example - Bad:

"This is an ancient Eastern breathing technique."

Protect Closed Practices​

Definition: Closed practices are those restricted to:

  • Initiated members of a tradition
  • Specific lineages
  • Cultural insiders
  • Those who have undergone training

Examples:

  • Indigenous sacred ceremonies
  • Initiation-only practices
  • Lineage-specific transmissions
  • Cultural-insider-only rituals

Policy:

  • Do not disclose closed practices publicly
  • If uncertain, consult cultural experts
  • Err on side of protection

Context Preservation​

Required:

  • Explain traditional context
  • Note cultural significance
  • Preserve meaning while adapting form
  • Acknowledge limitations of adaptation

Enforcement​

  • HIGH severity - blocks until corrected
  • Cultural Reviewer must approve
  • Community input welcomed

3. Evidence with Humility​

"May help" not "will cure."

Match Language to Evidence​

Use the Evidence Language Scale:

  • Strong Evidence → "Research demonstrates..."
  • Moderate Evidence → "Studies suggest..."
  • Preliminary Evidence → "Early research indicates..."
  • Traditional Only → "Traditionally used for..."
  • Unknown → "The mechanism is not understood..."

Acknowledge Limitations​

Required in all evidence summaries:

  • Sample sizes
  • Study quality
  • Conflicting findings
  • "Results vary" language
  • Mechanism unknowns

Distinguish Correlation from Causation​

Good:

"Studies show an association between meditation and reduced anxiety."

Bad:

"Meditation eliminates anxiety."

Enforcement​

  • HIGH severity for overclaims
  • Clinical Reviewer must approve
  • Evidence must be cited (PMID/DOI)

4. Empower, Don't Control​

User autonomy is sacred.

Provide Information, Don't Prescribe​

Empowering:

"You might consider trying this practice. Notice how it feels for you."

Controlling:

"You must do this practice daily or you won't heal."

No Coercive Language​

Avoid:

  • "You have to..."
  • "You must..."
  • "The only way..."
  • "If you don't..."

Prefer:

  • "You might..."
  • "One option is..."
  • "Some people find..."
  • "If this resonates..."

User Controls Data​

Required:

  • All health data stays local
  • User can export/delete anytime
  • No required accounts
  • Clear about what's collected

Clear About Limitations​

Required:

  • "This is not medical advice"
  • "Consult professionals"
  • "One tool among many"
  • "Results vary"

Enforcement​

  • MEDIUM severity for prescriptive language
  • HIGH severity for coercion
  • User data control is CRITICAL

5. Privacy as Sanctuary​

Healing data stays with the healer.

Local-First Architecture​

Required:

  • All user health data stored locally
  • No external transmission without explicit consent
  • No centralized health records
  • User controls everything

No Tracking​

Prohibited:

  • Analytics without consent
  • Third-party cookies
  • Fingerprinting
  • PII collection

Allowed (with opt-in):

  • Anonymous aggregate analytics
  • Crash reporting (no PII)
  • Usage patterns (anonymized)

No Required Accounts​

Policy:

  • Core functionality works without account
  • Optional accounts for sync only
  • No email required
  • No social login pressure

Data Minimization​

Principle:

  • Collect only what's necessary
  • Store only what's useful
  • Delete what's not needed
  • User can delete everything

Enforcement​

  • CRITICAL severity - absolute requirement
  • Technical architecture must support
  • Regular privacy audits

6. Accessible to All​

Design for the most vulnerable users.

No Cost Gatekeeping​

Policy:

  • Essential content is free
  • No paywalls for safety information
  • Optional premium features only
  • Open source preferred

WCAG Compliance​

Required:

  • WCAG AA minimum (AAA preferred)
  • Keyboard navigation
  • Screen reader compatible
  • Color contrast sufficient
  • No color-only information

Motion Sensitivity​

Required:

  • Respect prefers-reduced-motion
  • Provide static alternatives
  • No auto-play
  • User controls animations

One-Handed Operation​

Required:

  • All features work one-handed
  • Buttons appropriately sized
  • No complex gestures required
  • Voice control supported

Works Offline​

Required:

  • Core functionality offline
  • Graceful degradation
  • Local data storage
  • Clear online/offline status

Enforcement​

  • HIGH severity for accessibility issues
  • Accessibility Auditor must approve
  • Regular compliance testing

7. Continuous Improvement​

Learn from outcomes, update with evidence.

Regular Ethics Audits​

Frequency: Annual minimum

Reviews:

  • All content for accuracy
  • All claims for evidence support
  • All attributions for completeness
  • All disclaimers for presence

Feedback Loops​

Sources:

  • User reports
  • Practitioner feedback
  • Community input
  • Incident reviews

Response:

  • Acknowledge receipt
  • Investigate thoroughly
  • Update if warranted
  • Communicate changes

Stay Current with Research​

Process:

  • Monitor relevant journals
  • Update evidence summaries
  • Revise language when appropriate
  • Note when studies contradict

Evolve with Understanding​

Principle:

  • No content is final
  • Evidence evolves
  • Cultural understanding deepens
  • Technology improves

Action:

  • Version all content
  • Track changes
  • Document rationale
  • Enable rollback

Enforcement​

  • Regular review schedule maintained
  • Updates logged and audited
  • Community input welcomed

Principle Conflicts​

Occasionally, principles may appear to conflict. Resolution:

Hierarchy​

When principles conflict:

  1. First, Do No Harm - Always wins
  2. Honor Traditions - Second priority
  3. Privacy - Third priority
  4. Others - Case by case

Example​

Scenario: A traditional practice might cause harm.

Resolution:

  1. Harm prevention takes priority
  2. Document why practice excluded
  3. Explain to tradition holders
  4. Seek safer adaptations

Escalation​

When unclear:

  1. Ethics Guardian reviews
  2. Human expert consulted
  3. Community input sought
  4. Decision documented

"These principles are not suggestions. They are requirements. They protect the vulnerable. They honor the sacred. They guide every action."