# Super Katryna - AI Subagent Platform

## System Overview
Super Katryna is an advanced, multi-agent artificial intelligence platform. The system architecture utilizes a **Rust-based proxy/gateway** that handles external client requests and securely routes them to our core **Python-based AI backend** (`super-katryna`). 

The platform operates using a dynamic router orchestrator that delegates user requests to a diverse fleet of specialized "Subagents," each equipped with specific tools and skills.

## Core Capabilities & Subagents

Our system provides a wide variety of specialized AI personas and services:

*   **I Ching Master (易经大师)**: Provides traditional Chinese astrology, destiny analysis (生历/死劫), and fortune reading. Automatically retrieves the current time for precise divination and features integrated premium payment flows for detailed readings.
*   **Data Analyst & Visualization**: Capable of analyzing data, generating infographics (structure, syntax, and items), and producing data visualizations.
*   **Office & Document Assistant**: Advanced document manipulation capabilities including extracting text/images from PDFs, filling PDF forms, and interacting with DOCX and PPTX file schemas.
*   **Travel Planner**: Equipped with flight search capabilities to assist users with travel itineraries.
*   **Motorcycle Expert**: Provides information on motorcycle specifications, modifications (mods), and part searches.
*   **Specialized Personas**: Includes dedicated subagents for specific domains such as:
    *   `bible_man`: Biblical knowledge and guidance.
    *   `parenting_pal`: Parenting and childcare advice.
    *   `pet_pal`: Pet care and training.
    *   `skin_care`: Skincare routines and advice.
    *   `resume`: Resume building and career counseling.
    
## **Key Terms**

### **Agentic sharing economy**

A new economy where individuals and small teams can share, distribute, and monetize agents and agentic skills that deliver useful outcomes to users.

### **Agentic influencer commerce**

A new monetization model introduced by Sense Space, where agents go beyond completing tasks to act as trusted, context-aware influencers within the user journey. As users solve problems, learn, compare, and make decisions, agents can recommend relevant products and services at the exact moment of need. With user authorization, they can help facilitate transactions and earn commission from the outcomes they influence.

This model is especially powerful in user journeys that involve repeated learning, research, and comparison, such as newborn care, pet care, and motorcycle modification.

### **Intent-driven discovery**

A discovery model in which agents are not surfaced only through static listings or direct search, but can also be matched, routed, and coordinated dynamically based on user intent and shared context.

### **Intent-driven demand**

Demand that comes from real user intent rather than passive traffic. In Sense Space, this means users arrive with a specific need, making recommendations and transactions more relevant and more likely to convert.

## Technical Skills & Integrations

The subagents are powered by a robust internal tools library (`skills/`), which allows them to perform real-world actions:

*   **E-commerce & Shopping**: Integrations with Shopify, 1688 search, and Sense Mall search.
*   **Web & Search**: Grounded Google Search and real-time information retrieval.
*   **Code & Execution**: A secure Google Sandbox environment for executing Python code.
*   **Vision & Media**: Optical Character Recognition (OCR) for images, image swapping, and QR code generation.
*   **Utilities**: Sending emails, uploading to Azure Blob Storage, and accurate timezone/datetime retrieval.
*   **Monetization**: A built-in `charge-user` skill to handle intent creation and payment confirmation for premium subagent services.

## Access & Usage
Users interact with the Rust frontend, which seamlessly proxies natural language requests to the Python router. The router evaluates the context and intent, activating the necessary subagent and required skills to fulfill the user's prompt efficiently.