Canonical source: the canonical features document for this domain is
deep-dive/features.md, rendered in the docs
center. This file is its operational companion (service topology, APIs,
schemas, runtime detail) — reconciled by linking, not duplicated (§2, §13).
Compassionate guardianship platform: predictive harm prevention, mindful
friction, performer protection, community harmony, rehabilitation pathways,
merit and karma systems, dharma analytics, and cross-domain ethics
integration.
Table of Contents#
- Foundation and Harm Classification
- Precognition -- Predictive Harm Detection
- Mindful Friction
- Community Harmony
- Performer Protection
- Rehabilitation Pathways
- Merit and Karma Systems
- Dharma Analytics
- Cross-Domain Ethics Integration
- Data Foundation
1. Foundation and Harm Classification#
1.1 Buddha Nature Framework#
| Capability |
Description |
| Innate goodness modeling |
Every user entity carries a BuddhaNature interface with numeric properties for compassion seed, wisdom factor, mindfulness level, karma balance, innate goodness, and awakening potential |
| Refuge status tracking |
Five-stage spiritual progress model: seeking, taking refuge, established, deepening, realized |
| Compassion-first moderation |
All moderation actions guided by the principle that users carry innate potential for wholesome action |
| Growth orientation |
Moderation designed to guide toward wholesome expression rather than punish unwholesome behavior |
1.2 Harm Category Taxonomy#
| Capability |
Description |
| 50+ harm categories |
Comprehensive catalog of harmful content and behavior types organized into 10 domains |
| Interpersonal harm |
Targeted harassment, repeated harassment, sexual harassment, bullying, threats of violence, intimidation |
| Identity-based harm |
Racial hate speech, ethnic hate speech, religious hate speech, gender-based hate speech, homophobia, transphobia, ableism, xenophobia |
| Safety harm |
CSAM (child sexual abuse material), grooming, self-harm promotion, suicide promotion, violence promotion, terrorism promotion |
| Integrity harm |
Misinformation, disinformation, impersonation, fraud, intellectual property theft |
| Exploitation harm |
Sexual exploitation, labor exploitation, financial exploitation, emotional manipulation |
| Manipulation harm |
Social engineering, coordinated inauthentic behavior, astroturfing, vote manipulation |
| Disruption harm |
Spam, flooding, vandalism, trolling, ban evasion |
| Legal harm |
Defamation, doxxing (publishing private information), blackmail, extortion, illegal content distribution |
| Platform harm |
Terms of service violations, abuse of reporting systems, ban evasion, multi-accounting |
| Systemic harm |
Algorithmic bias, discriminatory patterns, structural inequality reinforcement |
1.3 Severity Classification#
| Capability |
Description |
| 6-level severity scale |
Negligible, low, moderate, significant, high, and critical severity classifications |
| Context-sensitive severity |
Severity assessment considers context, intent, audience, repetition, and impact |
| Escalation thresholds |
Clear thresholds for when automated responses escalate to human review |
| Severity-based routing |
Different intervention types triggered at different severity levels |
1.4 Foundation Utilities#
| Capability |
Description |
| Platform constants |
Standardized moderation constants for consistent enforcement across the platform |
| Shared utilities |
Common utility functions for content analysis, severity scoring, and harm assessment |
| Domain error types |
Specialized error types for moderation failures, classification errors, and intervention exceptions |
2. Precognition -- Predictive Harm Detection#
2.1 Intent Analysis#
| Capability |
Description |
| Message intent classification |
Analyze the underlying intent behind user messages before they are published |
| Harmful intent detection |
Identify messages crafted to harass, deceive, manipulate, or harm |
| Benign intent recognition |
Distinguish genuinely harmful messages from sarcasm, jokes, reclaimed language, and cultural expression |
| Intent confidence scoring |
Provide confidence scores for intent classifications to avoid false positives |
| Multi-intent support |
Recognize when a single message contains mixed intents (partly helpful, partly harmful) |
2.2 Emotional Detection#
| Capability |
Description |
| Distress signal recognition |
Detect signs of emotional distress in user content (despair, crisis, overwhelm) |
| Anger escalation detection |
Identify escalating anger patterns before they result in harmful outbursts |
| Emotional state modeling |
Build a real-time model of user emotional state from content and behavior signals |
| Emotional contagion tracking |
Monitor when negative emotional states spread through community interactions |
| Supportive intervention triggers |
Automatically trigger supportive resources when distress signals are detected |
2.3 Typing Dynamics Analysis#
| Capability |
Description |
| Typing speed monitoring |
Analyze changes in typing speed that may indicate emotional escalation |
| Deletion pattern detection |
Detect excessive deletion and rewriting as a signal of frustration or self-censoring |
| Draft analysis |
Analyze drafted-but-not-yet-sent content for harmful patterns |
| Cooling-off indicators |
Identify when typing dynamics suggest the user is calming down versus escalating |
2.4 Behavioral Pattern Tracking#
| Capability |
Description |
| Historical behavior profiling |
Build behavioral profiles from interaction history to identify concerning trends |
| Pattern deviation alerts |
Alert when user behavior deviates significantly from their established baseline |
| Escalation trajectories |
Identify users on escalating trajectories before they cross harm thresholds |
| Positive behavior recognition |
Also track positive behavioral patterns to reinforce constructive engagement |
| Recidivism prediction |
Predict likelihood of repeat violations based on behavioral history and intervention response |
2.5 Context Awareness#
| Capability |
Description |
| Conversation context |
Understand messages within the full context of the ongoing conversation |
| Relationship context |
Consider the relationship between users (strangers, friends, partners) when assessing potential harm |
| Community context |
Factor in community norms and culture when evaluating content |
| Temporal context |
Consider time of day, event context, and situational factors in harm assessment |
| Platform context |
Adapt detection thresholds to the specific platform and content type |
2.6 Cascade Prediction#
| Capability |
Description |
| Viral harm prediction |
Predict when a single harmful message could cascade into broader community harm |
| Pile-on detection |
Detect early signs of coordinated pile-on behavior against a user |
| Misinformation spread modeling |
Model how false information might spread through community networks |
| Intervention timing optimization |
Determine the optimal moment to intervene for maximum harm reduction with minimum disruption |
| Cascade severity estimation |
Estimate the potential severity and reach of a harmful cascade before it fully develops |
3. Mindful Friction#
3.1 Pause and Breathe#
| Capability |
Description |
| Pre-posting pause |
Introduce a brief pause before harmful content is published, giving the user a moment to reconsider |
| Breathing exercise integration |
Offer an optional short breathing exercise during the pause period |
| Cool-down timers |
Configurable cool-down periods proportional to detected emotional escalation |
| Voluntary reflection |
Frame the pause as an invitation to reflect rather than a punishment or barrier |
| Bypass respect |
Allow users to proceed after the pause -- friction slows but never blocks expression |
3.2 Samma Vaca (Right Speech)#
| Capability |
Description |
| Right Speech framework |
Gentle prompts based on the Buddhist Noble Eightfold Path's Right Speech principles |
| Truthfulness check |
"Is what you are about to say true?" prompt for potentially misleading content |
| Helpfulness check |
"Is this helpful to the conversation?" prompt for potentially harmful content |
| Kindness check |
"Is this said with kindness?" prompt for potentially cruel or hurtful content |
| Timeliness check |
"Is this the right time and place?" prompt for potentially disruptive content |
| Non-judgmental framing |
All prompts are presented as gentle invitations to reflection, never accusations |
3.3 Cognitive Reframing#
| Capability |
Description |
| Automatic thought identification |
Help users recognize when their message is driven by a cognitive distortion |
| Distortion labeling |
Identify specific cognitive distortions (all-or-nothing thinking, catastrophizing, personalizing, mind reading) |
| Reframing suggestions |
Offer alternative ways to interpret the situation that triggered the harmful response |
| Perspective broadening |
Encourage consideration of multiple perspectives before responding |
3.4 Perspective Shift#
| Capability |
Description |
| Empathy prompts |
"How might the other person feel reading this?" prompts before potentially hurtful messages |
| Role reversal |
"How would you feel if someone sent this to you?" perspective-taking exercises |
| Impact preview |
Show the potential emotional impact of a message on its recipient |
| Third-party perspective |
"How would a neutral observer view this exchange?" framing |
3.5 Compassion Nudges#
| Capability |
Description |
| Kindness reminders |
Gentle reminders of shared humanity during heated interactions |
| Positive intent assumptions |
Encourage assuming positive intent when messages could be interpreted multiple ways |
| Common ground highlighting |
Surface shared interests or values between conflicting users |
| Gratitude prompts |
Redirect attention toward what is positive about the community and relationships |
3.6 Alternative Expression#
| Capability |
Description |
| Rewrite suggestions |
Suggest alternative phrasings that express the same sentiment without harmful content |
| Tone adjustment |
Offer tone-shifted versions of messages (e.g., from aggressive to assertive, from passive-aggressive to direct) |
| Constructive criticism framing |
Help users express valid criticism constructively rather than destructively |
| Frustration channels |
Suggest appropriate outlets for frustration (private journaling, support channels, feedback forms) |
4.1 Temperature Monitor#
| Capability |
Description |
| Community emotional temperature |
Track aggregate emotional sentiment across the community in real time |
| Temperature trend tracking |
Monitor whether community mood is warming (positive), cooling (negative), or stable |
| Heat map visualization |
Visualize community sentiment geographically, temporally, and by topic |
| Early warning thresholds |
Alert moderators when community temperature approaches concerning levels |
| Temperature history |
Maintain historical temperature data for retrospective analysis |
4.2 Conflict Detection#
| Capability |
Description |
| Emerging conflict identification |
Detect conflicts between users in their early stages before escalation |
| Relationship tension mapping |
Track tension levels between specific user pairs over time |
| Topic-based conflict clustering |
Identify contentious topics that repeatedly generate conflict |
| Mediation triggers |
Automatically suggest mediation when conflicts reach predetermined intensity levels |
| De-escalation monitoring |
Track whether conflicts are de-escalating after intervention |
4.3 Raid Defense#
| Capability |
Description |
| Coordinated attack detection |
Identify coordinated harassment campaigns (raids, brigading) in real time |
| Automated response |
Activate defensive measures (rate limiting, temporary restrictions, increased moderation) during raids |
| Source tracking |
Trace raid origins to external coordination platforms |
| Pattern library |
Maintain a library of known raid patterns for faster future detection |
| Post-raid analysis |
Analyze raid events after the fact to improve future defenses |
4.4 Post-Incident Healing#
| Capability |
Description |
| Community acknowledgment |
Structured processes for community leaders to acknowledge harmful events |
| Healing spaces |
Create temporary moderated spaces for community members to process harmful events together |
| Resource provision |
Surface relevant support resources after traumatic community events |
| Recovery monitoring |
Track community recovery after major incidents through sentiment and engagement metrics |
| Lessons learned |
Document incidents and responses for continuous improvement |
4.5 Culture Cultivation#
| Capability |
Description |
| Positive norms reinforcement |
Proactively reinforce desired community norms through highlighting positive behavior |
| Community guidelines visibility |
Keep community guidelines visible and accessible with contextual reminders |
| Ambassador programs |
Tools to empower community members as positive culture ambassadors |
| Celebration of positive contributions |
Surface and celebrate members who contribute positively to community health |
| Norm evolution |
Facilitate democratic evolution of community norms as the community grows and changes |
5.1 Real-Time Shield#
| Capability |
Description |
| Live content filtering |
Filter harmful content from chat, comments, and messages in real time during live performances |
| Keyword and pattern blocking |
Block known harmful keywords, phrases, and patterns from reaching performers |
| Adaptive filtering |
Dynamically adjust filter sensitivity based on current threat level and performer preferences |
| Performer-defined rules |
Allow performers to set their own content filtering rules and boundaries |
| Filter transparency |
Performers can review what was blocked and adjust filters accordingly |
5.2 Parasocial Detection#
| Capability |
Description |
| Attachment pattern recognition |
Detect unhealthy parasocial attachment patterns in fan behavior |
| Obsessive behavior indicators |
Identify excessive messaging frequency, possessive language, and fantasy-blurring behavior |
| Escalation risk assessment |
Score parasocial relationships by escalation risk (low, moderate, high, critical) |
| Gentle intervention |
Provide resources and gentle guidance to fans exhibiting parasocial patterns before they reach harmful levels |
| Performer notification |
Alert performers to concerning parasocial patterns affecting them |
5.3 Boundary Enforcement#
| Capability |
Description |
| Boundary definition |
Allow performers to define clear interaction boundaries (topics, contact methods, response expectations) |
| Automated enforcement |
Automatically enforce defined boundaries without requiring performer intervention |
| Boundary communication |
Clearly communicate performer boundaries to fans in context |
| Boundary violation tracking |
Track boundary violations and escalate repeat offenders |
| Customizable responses |
Performers choose how boundary violations are handled (warning, redirect, block) |
5.4 NCII and Deepfake Protection#
| Capability |
Description |
| Non-consensual intimate imagery detection |
Detect and prevent distribution of non-consensual intimate images |
| Deepfake detection |
Identify AI-generated or manipulated images and videos using performers' likenesses |
| Proactive scanning |
Scan for performer likeness misuse across platform content |
| Takedown workflows |
Streamlined processes for requesting and executing content takedowns |
| Evidence preservation |
Preserve evidence of violations for legal and platform action |
| Capability |
Description |
| Mental health monitoring |
Track performer wellness indicators from interaction patterns and content |
| Burnout detection |
Identify signs of performer burnout from activity patterns and sentiment |
| Wellness check-ins |
Scheduled wellness check-ins with guided self-assessment |
| Resource provision |
Provide access to mental health resources, peer support, and professional services |
| Break encouragement |
Encourage healthy breaks based on detected fatigue or stress patterns |
| Capability |
Description |
| Safety metrics overview |
Aggregate view of all safety-related metrics for the performer's presence |
| Threat level indicators |
Current threat level assessment based on recent activity |
| Blocked content statistics |
Summary of content blocked by the real-time shield |
| Parasocial alerts |
Active parasocial concern notifications with recommended actions |
| Wellness score |
Overall performer wellness score with trend indicators |
| Historical analysis |
View safety and wellness trends over time to identify seasonal or content-related patterns |
6. Rehabilitation Pathways#
6.1 Shadow Work Journeys#
| Capability |
Description |
| Guided self-reflection |
Structured self-reflection journeys for users who have committed violations |
| Understanding impact |
Exercises to help violators understand the impact of their actions on others |
| Root cause exploration |
Guided exploration of the underlying causes of harmful behavior |
| Personal growth goals |
Set growth goals related to the specific violation type |
| Journey customization |
Journeys tailored to the specific type and severity of violation |
6.2 Journey Progress Tracking#
| Capability |
Description |
| Milestone tracking |
Track progress through rehabilitation milestones |
| Completion verification |
Verify genuine engagement with rehabilitation exercises (not just clicking through) |
| Progress visualization |
Visual representation of rehabilitation journey progress |
| Adaptive pacing |
Adjust journey pacing based on demonstrated engagement and understanding |
| Support check-ins |
Regular check-ins throughout the rehabilitation journey |
6.3 Empathy Training#
| Capability |
Description |
| Perspective-taking exercises |
Structured exercises to build empathy and perspective-taking skills |
| Victim impact understanding |
Help violators understand the full impact of their behavior on victims |
| Emotional literacy development |
Build emotional vocabulary and recognition skills |
| Empathy assessment |
Assess empathy development over the course of training |
| Real-world application |
Connect empathy training to real-world scenarios the user is likely to encounter |
6.4 Restorative Circles#
| Capability |
Description |
| Facilitated conversations |
Structured facilitation framework for conversations between violators and those they harmed |
| Safety protocols |
Comprehensive safety protocols to protect all participants in restorative processes |
| Agreement building |
Facilitate concrete agreements for repair and changed behavior |
| Circle preparation |
Prepare all participants before the restorative circle conversation |
| Follow-up tracking |
Monitor adherence to restorative agreements after the circle concludes |
6.5 Accountability Tracking#
| Capability |
Description |
| Commitment recording |
Record specific commitments made during rehabilitation and restorative processes |
| Compliance monitoring |
Track whether commitments are being honored over time |
| Positive behavior reinforcement |
Recognize and reinforce positive behavior changes |
| Accountability partnerships |
Connect rehabilitating users with volunteer accountability partners |
| Transparent reporting |
Provide transparency to affected parties about accountability progress |
6.6 Reintegration Pathways#
| Capability |
Description |
| Graduated privilege restoration |
Gradually restore platform privileges as rehabilitation progresses |
| Monitored reintegration |
Enhanced monitoring during the reintegration period |
| Community readiness assessment |
Assess whether the community is ready for the user's return |
| Reintegration support |
Provide support resources during the transition back to full participation |
| Relapse prevention |
Identify and address risk factors for repeated violations during reintegration |
7. Merit and Karma Systems#
7.1 Merit Accumulation#
| Capability |
Description |
| Positive action tracking |
Track positive contributions (helpful content, supportive interactions, community building) |
| Merit point earning |
Award merit points for verified positive actions |
| Category-specific merit |
Track merit in different categories (helpfulness, creativity, mentorship, moderation, knowledge sharing) |
| Diminishing returns |
Prevent merit gaming through diminishing returns on repetitive actions |
| Community-validated merit |
Allow community endorsement of particularly valuable contributions |
7.2 Merit Calculation#
| Capability |
Description |
| Multi-factor scoring |
Calculate merit scores from multiple factors (quality, impact, consistency, community validation) |
| Weighted calculation |
Different action types weighted according to their positive impact |
| Decay prevention |
Ensure merit does not decay unfairly over time while still rewarding recent activity |
| Anti-gaming measures |
Algorithms to detect and prevent merit manipulation |
7.3 Privilege Tiers#
| Capability |
Description |
| Merit-based access levels |
Unlock additional platform privileges as merit accumulates |
| Graduated trust |
Higher-merit users receive more trust and fewer friction interventions |
| Tier transparency |
Clear communication about what each privilege tier provides and how to reach it |
| Tier maintenance |
Require ongoing positive behavior to maintain elevated privilege tiers |
| Tier reduction |
Reduce privileges when behavior degrades, with clear explanation and recovery path |
7.4 Karma Visibility#
| Capability |
Description |
| Personal karma dashboard |
Users can view their own karma scores and contributing factors |
| Public karma indicators |
Optional public display of karma level (configurable visibility) |
| Karma transparency |
Clear explanation of how karma scores are calculated |
| Privacy controls |
Granular control over what karma information is visible to others |
7.5 Achievement System#
| Capability |
Description |
| Positive community achievements |
Achievements for constructive community participation milestones |
| Mentorship achievements |
Achievements for helping new users integrate into the community |
| Consistency achievements |
Achievements for sustained positive behavior over time |
| Recovery achievements |
Achievements for successfully completing rehabilitation pathways |
8. Dharma Analytics#
| Capability |
Description |
| Aggregate health dashboards |
Real-time dashboards showing overall community health across multiple dimensions |
| Toxicity metrics |
Track harmful content rates, severity distribution, and trends over time |
| Positive engagement ratios |
Measure the ratio of positive to negative interactions |
| Community growth health |
Assess whether community growth is healthy (balanced) or problematic (influx of bad actors) |
| Comparative benchmarking |
Compare community health metrics against historical baselines and platform averages |
8.2 Individual Dharma Path#
| Capability |
Description |
| Personal growth tracking |
Track each user's growth along their individual dharma (purpose/path) |
| Behavioral improvement visualization |
Visualize how individual behavior has improved over time |
| Strength identification |
Identify areas where users are particularly positive community contributors |
| Growth opportunity identification |
Identify areas where users could develop more positive engagement patterns |
| Path recommendations |
Suggest activities and practices to support continued personal growth |
8.3 Predictive Wellness#
| Capability |
Description |
| Community wellness forecasting |
Predict future community wellness trends based on current indicators |
| Risk factor identification |
Identify factors that predict declining community health |
| Intervention effectiveness prediction |
Predict which interventions are most likely to be effective for specific situations |
| Early warning models |
Machine learning models that detect early signs of community health degradation |
| Seasonal pattern recognition |
Identify seasonal and cyclical patterns in community wellness |
8.4 Wisdom Reports#
| Capability |
Description |
| Automated insight generation |
Generate human-readable reports summarizing community health and trends |
| Leadership briefings |
Periodic reports for community leaders with actionable recommendations |
| Trend narratives |
Narrative explanations of significant trends and their likely causes |
| Success stories |
Highlight successful rehabilitation stories and positive community developments |
| Pattern documentation |
Document recurring patterns for institutional knowledge building |
8.5 Moderation Transparency#
| Capability |
Description |
| Transparency reports |
Regular public reports on moderation actions, volumes, and outcomes |
| Action accounting |
Detailed accounting of all moderation actions by type, severity, and outcome |
| Appeal outcomes |
Report on appeal rates, success rates, and overturned decisions |
| Algorithm transparency |
Clear documentation of how automated moderation decisions are made |
| Fairness auditing |
Regular audits for bias in moderation actions across user demographics |
9. Cross-Domain Ethics Integration#
9.1 Aphrodite Integration (Relationships/Dating)#
| Capability |
Description |
| Dating safety |
Apply Kuanyin harm detection to relationship and dating contexts |
| Consent verification |
Ethics frameworks for consent management in intimate interactions |
| Harassment prevention |
Specialized harassment detection tuned for romantic and social contexts |
| Boundary communication |
Support clear boundary communication in relationship settings |
| Exploitation prevention |
Detect patterns of emotional or financial exploitation in relationships |
9.2 Lilith Integration (Consciousness/Adult Content)#
| Capability |
Description |
| Content safety classification |
Apply appropriate content safety standards to consciousness and adult content domains |
| Performer protection overlay |
Extend performer protection capabilities to the Lilith domain |
| Age verification support |
Ethics framework for age verification and minor protection |
| Consent management |
Comprehensive consent tracking and verification for content creation and sharing |
| Exploitation detection |
Detect exploitation patterns specific to adult content platforms |
9.3 Hathor Integration (Worldbuilding)#
| Capability |
Description |
| Creative content moderation |
Adapt moderation for creative and fictional content contexts |
| Fiction versus reality distinction |
Appropriately handle harmful themes in clearly fictional/creative contexts |
| Collaborative safety |
Ensure safety in multi-user collaborative worldbuilding environments |
| Content rating |
Age-appropriate content rating for worldbuilding content |
9.4 Yemaya Integration (Creative Studio)#
| Capability |
Description |
| Creative safety |
Safety standards for creative tools and collaborative creation |
| Copyright and attribution |
Ethics framework for intellectual property and attribution in creative work |
| Creative community moderation |
Moderation tailored to creative feedback and critique contexts |
| Safe sharing |
Ensure safe content sharing and publication workflows |
| Capability |
Description |
| Unified ethics framework |
Consistent ethical standards applied across all Oshun platform domains |
| Cross-domain policy enforcement |
Enforce platform-wide policies while respecting domain-specific norms |
| Universal harm taxonomy |
Shared harm categories and severity levels across all domains |
| Cross-domain user reputation |
User reputation that follows them across domains for consistent trust assessment |
| Centralized reporting |
Unified reporting and transparency across all domain integrations |
10. Data Foundation#
10.1 Database Layer#
| Capability |
Description |
| Schema definitions |
Comprehensive database schemas for all moderation state (harm classifications, user moderation history, intervention records, merit scores) |
| Migration management |
Versioned database migrations for schema evolution |
| Data repositories |
Specialized data access repositories for each subdomain |
| Seed data |
Development and testing seed data for realistic development environments |
10.2 AI/ML Models#
| Capability |
Description |
| Intent classification models |
Machine learning models for classifying user intent |
| Harm detection models |
Neural network models trained on harm classification tasks |
| Behavioral prediction models |
Models for predicting user behavior based on historical patterns |
| Emotional state models |
Models for detecting and tracking emotional states from text and behavior |
10.3 Design Principles#
| Capability |
Description |
| Compassion over punishment |
All systems designed to guide rather than punish -- rehabilitation preferred over exclusion |
| Precognition over reaction |
Proactive detection and gentle intervention before harm occurs rather than after |
| User agency |
Mindful friction slows but never blocks -- users always maintain agency over their expression |
| Transparency |
All moderation decisions are explainable, auditable, and appealable |
| Growth orientation |
Every user is treated as capable of positive growth, regardless of past behavior |
| Cultural sensitivity |
Moderation adapts to cultural context, avoiding false positives from cultural misunderstanding |