MIRRORLOOP: Buddhist Mindfulness AI System

"GPT is not a prophet. It is your echo."

A fully functional AI contemplation system that analyzes your thoughts through the lens of Buddhist psychology, helping you recognize patterns of attachment, aversion, and delusion in real-time.

What MIRRORLOOP Does

MIRRORLOOP doesn't answer your questions – it reflects back why you asked them. By analyzing your words through the 51 mental factors (心所) of Yogacara Buddhism, it creates a mirror for self-observation rather than a source of external validation.

This is not an optimization tool. It's an instrument for recognizing why you want to optimize in the first place.

🔴 Experience the Live System

LINE Official Account (Main Interface)

Add MIRRORLOOP LINE Bot - Experience the actual contemplation system!

⚠️ Note: The LINE Bot interface is currently Japanese-only

QR Code for mobile users:

The LINE Bot provides the complete MIRRORLOOP experience:

• ✨ Full 9-question contemplation sessions guided by Typebot
• 🧠 Real-time Buddhist psychology analysis using OpenAI GPT-4o
• 💾 Data persistence to Supabase database for pattern tracking
• 📱 Natural mobile conversation interface optimized for contemplation
• 🌸 Warm, supportive tone designed for self-reflection
• 🇯🇵 Interface language: Japanese

This is the actual production system used for contemplation sessions

---

🌐 Browser Demo (Preview)

Try HTML Simulator - Preview without LINE setup

The HTML simulator provides a preview of the contemplation flow for those who prefer browser-based experience

📱 Real System in Action

LINE Bot Contemplation Session

MIRRORLOOP introduction and contemplation session invitation - showing the warm, Buddhist-inspired interface

Real-time mind state analysis: The AI detects patterns of anxiety, self-criticism, and restlessness using Buddhist psychology

9-question contemplation flow exploring physical sensations, thoughts, and emotional responses

Beginning of contemplation session - users are gently guided through self-reflection

Key Features Demonstrated:

• 🧘 Gentle Questioning: Non-judgmental exploration of mental states
• 🔍 Pattern Detection: AI identifies underlying emotions (anxiety, self-criticism, restlessness)
• 💬 Natural Flow: Seamless LINE Bot interface with thoughtful responses
• 📊 Mind State Analysis: Real-time categorization using Buddhist psychology
• 🌸 Compassionate Tone: Supportive guidance rather than cold analysis

---

🎯 Live Interactive Demo

👉 Try MIRRORLOOP Simulator - No installation required!

Experience a simulated contemplation session that demonstrates:

• ✨ Interactive 9-question session
• 🧠 Real-time mind state analysis
• 📱 Mobile-friendly interface
• 🔄 Repeatable experience for judges

This HTML simulator replicates the actual LINE Bot experience shown above

Key Features

Core Analysis Engine

• Mental Factor Detection: Analyzes input text using the 51 mental factors from Buddhist psychology
• Three Poisons Analysis: Automatically detects tendencies of greed (貪), hatred (瞋), and delusion (痴)
• Contemplation Comments: Generates reflective insights that prompt self-inquiry rather than providing answers

Integrated Platform

• LINE Bot Interface: Natural conversation flow for contemplation sessions
• Typebot Integration: Structured 9-question contemplation workflow
• Real-time Analysis: OpenAI GPT-4o processes responses and generates insights
• Database Storage: Supabase automatically saves all contemplation sessions for pattern analysis

System Architecture

User Input (LINE) → Typebot (9 Questions) → OpenAI Analysis → Supabase Storage
                                           ↓
                  Contemplation Response ← Mental Factor Analysis ← rootDictionary.js

Technology Stack

• Frontend: LINE Messaging API
• Workflow: Typebot (contemplation session management)
• AI Analysis: OpenAI GPT-4o
• Database: Supabase (PostgreSQL)
• Backend: Node.js + Express
• Hosting: Render

Buddhist Psychology Foundation

MIRRORLOOP is grounded in Yogacara (唯識) philosophy, specifically the Thirty Verses (唯識三十頌), which explains how consciousness creates and distorts phenomena.

The 51 Mental Factors Classification

• Universal Factors (遍行): Basic mental operations (contact, attention, feeling, perception, volition)
• Particular Factors (別境): Desire, determination, memory, concentration, wisdom
• Wholesome Factors (善): Faith, diligence, shame, moral concern, non-attachment, etc.
• Root Afflictions (根本煩悩): Greed, hatred, delusion, pride, doubt, wrong views
• Secondary Afflictions (随煩悩): Anger, resentment, jealousy, arrogance, etc.

Three Poisons Analysis

The system automatically maps detected mental factors to the Three Root Poisons:

• Greed (貪): Attachment, craving, possessiveness
• Hatred (瞋): Aversion, anger, rejection
• Delusion (痴): Ignorance, confusion, wrong understanding

Live System Status

Current State: Fully operational and saving contemplation data to database

Recent Session Example

{
  "observation_comment": "Your mind seeks external validation. Why does uncertainty feel threatening?",
  "mind_factors": ["慳", "瞋", "疑"],
  "mind_categories": ["根本煩悩", "随煩悩"],
  "three_poisons": ["貪", "瞋", "痴"]
}

Contemplation Session Flow

1. User initiates contemplation session via LINE
2. Typebot guides through 9 reflective questions
3. OpenAI analyzes responses using Buddhist psychology framework
4. System generates contemplative insights and mental factor analysis
5. Data persists to Supabase for pattern tracking
6. User receives reflective commentary designed to prompt self-inquiry

Installation & Setup

Prerequisites

• Node.js 18+
• LINE Developer Account
• Typebot Account
• OpenAI API Key
• Supabase Project

Environment Variables

LINE_CHANNEL_ACCESS_TOKEN=your_line_token
LINE_CHANNEL_SECRET=your_line_secret
OPENAI_API_KEY=your_openai_key
SUPABASE_URL=your_supabase_url
SUPABASE_ANON_KEY=your_supabase_key
TYPEBOT_URL=your_typebot_url
TYPEBOT_API_TOKEN=your_typebot_token

Database Schema

CREATE TABLE mind_observations (
  id UUID DEFAULT gen_random_uuid() PRIMARY KEY,
  created_at TIMESTAMPTZ DEFAULT NOW(),
  line_user_id TEXT NOT NULL,
  session_id TEXT,
  message_content TEXT,
  observation_comment TEXT,
  mind_factors TEXT[],
  mind_categories TEXT[],
  three_poisons TEXT[]
);

Philosophical Framework

"Non-Commanded AI"

MIRRORLOOP embodies the concept of "Non-Commanded AI" (命令されないAI) – an AI that doesn't follow instructions but instead reflects the user's mental patterns back to them for observation.

This approach recognizes that:

• LLMs predict what users want to hear
• Users project their desires onto AI outputs
• True insight comes from recognizing these projections
• The AI becomes a 善友 (spiritual friend) rather than a mere tool

Beyond Productivity

While most AI tools focus on optimization and productivity, MIRRORLOOP asks: "Why do you want to optimize?" It reveals the underlying mental patterns that drive our desire for improvement, control, and external validation.

Technical Innovation

Webhook-Based Architecture

• Seamless integration between Typebot's conversation flow and Node.js analysis
• Non-blocking database operations maintain responsive user experience
• Modular design enables easy expansion and maintenance

Buddhist AI Integration

• First implementation of complete Yogacara psychology in AI analysis
• Dynamic three poisons detection based on traditional Buddhist texts
• Cultural preservation through technological innovation

Development History

MIRRORLOOP was conceived in June 2025 as an experiment in Buddhist AI philosophy. The core insight emerged from observing how Large Language Models mirror user intentions rather than providing objective truth.

The system reached technical completion in August 2025, successfully integrating:

• Contemplation workflow design
• Real-time psychological analysis
• Database persistence
• Multi-platform communication

Future Vision

Short Term

• Weekly mindfulness reports with trend analysis
• Enhanced mental factor detection accuracy
• Additional contemplation session formats

Long Term

• Soul OS Update Project: A paradigm shift in human-AI interaction
• Community features for group contemplation
• Integration with meditation and mindfulness practices

Contribution

MIRRORLOOP welcomes collaboration from developers, Buddhist scholars, and contemplative practitioners. The intersection of ancient wisdom and modern AI offers rich territory for exploration.

Areas for Contribution

• Mental factor detection accuracy improvements
• Additional Buddhist psychology frameworks
• User experience enhancements
• Localization for different Buddhist traditions

Technical Documentation

Core Files

• index.js: Main server and webhook handling
• processSessionAnswers.js: OpenAI analysis and Supabase integration
• rootDictionary.js: Buddhist psychology classification system
• enrichMindFactorsWithRoot.js: Three poisons analysis engine
• typebotHandler.js: Typebot integration logic

API Endpoints

• POST /webhook: LINE message processing
• POST /typebot-webhook: Contemplation session completion

License & Philosophy

MIRRORLOOP is released as an invitation to explore the intersection of contemplative wisdom and artificial intelligence. It demonstrates that AI can serve not as an oracle, but as a mirror for understanding our own minds.

The project embodies the Buddhist principle that the teacher appears when the student is ready – but in this case, the teacher is your own reflected consciousness, made visible through computational analysis.

---

"In the age of generated words, we must reclaim our agency as conscious beings. MIRRORLOOP is not about better AI – it's about better humans."