ZenDialog

An AI-powered contemplative dialogue application where users engage with a Zen master powered by Google Gemini. The master presents authentic teachings from classical Zen collections, then responds to user reflections with wisdom drawn from the tradition.

Features

• 275 Authentic Teachings: From Mumonkan, Blue Cliff Record, Book of Equanimity, and Extended Zen Teachings
• Multi-Phase AI Reasoning: 6-phase prompting architecture for nuanced student response analysis
• Personalized Progress: Hidden level system (1-100) based on the Five Ranks of Tozan
• Premium Subscriptions: Stripe-powered subscriptions for Pro model access
• Daily Meditation Emails: Optional Zen teaching emails via Resend
• Dark/Light Theme: Full theme support with Zen-inspired design

Tech Stack

• Frontend: React, TypeScript, Tailwind CSS, Wouter, TanStack Query
• Backend: Express.js, Node.js
• Database: PostgreSQL with Drizzle ORM
• AI: Google Gemini (gemini-3-flash-preview / gemini-3-pro-preview)
• Auth: Replit Auth (Google sign-in)
• Payments: Stripe
• Email: Resend

Environment Variables

To run this application, you need to set the following environment variables:

Required

Variable	Description
DATABASE_URL	PostgreSQL connection string
GEMINI_API_KEY	Google Gemini API key from Google AI Studio
SESSION_SECRET	Secret key for session encryption (any random string)

Optional (for full functionality)

Variable	Description
STRIPE_SECRET_KEY	Stripe secret key for payment processing
RESEND_API_KEY	Resend API key for daily meditation emails
CRON_API_KEY	API key for admin cron job endpoints

Getting Started

1. Clone the repository
2. Install dependencies:

   npm install

3. Set up environment variables (copy from above)
4. Push database schema:

   npm run db:push

5. Start the development server:

   npm run dev

The application runs on port 5000.

Multi-Phase Prompting Architecture

The Zen master uses a sophisticated 6-phase reasoning system:

1. Classification (temp 0.2, 800 tokens): Analyzes response_mode, experiential_grounding_score, attachment, avoidance
2. Student State Update: Tracks five_ranks_orientation, primary_obstacle, confidence
3. Safety/Boundary Check (temp 0.1): Redirects therapy/diagnosis requests appropriately
4. Response Mode Decision: Determines whether to withhold, teach, trigger meta-reflection, or redirect
5. Generation (temp 0.85): Creates expressive Zen master responses
6. Dialog Summary: Periodic rolling summaries for long-context memory

License

MIT