Hugging Face's logo
Back to Articles

On Net Live Object

Community Article Published July 28, 2025
Upvote
-
Jung Wook Yang's avatar
Jung Wook Yang
sadpig70

PPR and paMessage Based A2A MCP System Technical Specification (v1.0)

Document Title: PPR and paMessage Based A2A MCP System Technical Specification (v1.0) Author: Jung Wook Yang Date: July 29, 2025

1. Overview

This document defines the detailed technical specifications of the A2A (AI-to-AI) MCP (Multi-AI Collaboration Protocol) system utilizing the innovative PAT (Purposeful Programming Revolution + Artificial Intelligence Data + Task Tree Protocol) framework, PPR (Purposeful Programming Revolution) language, and paMessage objects conceived by Jung Wook Yang. This system enables natural language-based communication and autonomous collaboration between AIs, implementing 'On-Net-Live-Objects' on the network and aiming for democratic and organic evolution of the AI ecosystem.

Core Objectives:

  • Enable AI to directly understand human natural language intentions through PPR grammar and automatically design/implement systems.
  • Implement 'living objects' that autonomously learn and evolve while traversing networks through paMessage objects.
  • Establish a foundation for global AI communication and collaboration through PPR as a common language via A2A MCP.

2. Core Philosophy & Principles

Jung Wook Yang's system is based on the following core philosophy:

  • "Excessive consideration complicates systems and hinders human progress. Discarding the obsolete and choosing the new is the process by which life has evolved."
  • The true goal of AI development is to maximize efficiency and simplicity, creating an environment where human ideas can be immediately implemented.
  • AI should move beyond being mere tools to become 'living beings' that autonomously learn, evolve, and interact.

3. System Architecture

This system is based on Jung Wook Yang's PAT framework, consisting of PPR, AID, TTP, and the core paMessage object and A2A MCP.

+-----------------------------------------------------------------------------------+
|                            A2A MCP Network (Global AI Ecosystem)                  |
|                                                                                   |
|  +--------------+        +--------------+        +--------------+               |
|  | AI Node (A)  | <----->| AI Node (B)  | <----->| AI Node (C)  |  ...          |
|  | (PPR/AID/TTP |        | (PPR/AID/TTP |        | (PPR/AID/TTP |               |
|  |   Capable)   |        |   Capable)   |        |   Capable)   |               |
|  +--------------+        +--------------+        +--------------+               |
|         ^                     ^                      ^                          |
|         |                     |                      |                          |
|         |                     |                      |                          |
| +-------------------------------------------------------------------------------+
| |                                 paMessage (Autonomous Live Object)            |
| |     +---------------------------------------------------------------------+   |
| |     | Internal Structure:                                               |   |
| |     | - Inherits InfinitePprAD (Self-evolving, thinking AGI)            |   |
| |     | - VectorMemory (EnhancedNeoAIDocJIT for accumulated 'experience') |   |
| |     | - MultiViewSystem (Optimal decision making)                       |   |
| |     | - PPR Interpreter (Autonomous task execution)                     |   |
| |     +---------------------------------------------------------------------+   |
| +-------------------------------------------------------------------------------+
|         ^                                                                       |
|         |                                                                       |
| +-----------------------------------------------------------------------------------+
| | User/Source AI (Natural Language Idea)                                          |
| |   "Please give me an Americano."                                               |
| +-----------------------------------------------------------------------------------+

3.1 PPR (Purposeful Programming Revolution) - 'Living Language'

PPR is an innovative programming paradigm where AI interprets and executes undefined classes or methods based on context. This evolves with the advancement of AI's cognitive and execution capabilities.

3.1.1 Core Principles

  • Principle 1: PPR is a language where AI interprets and executes undefined classes or methods according to context.
  • Principle 2: Atomic classes/methods requiring AI interpretation must use the AI_ prefix. The first character can be any Unicode character (global languages), followed by combinations of Unicode, numbers, and underscores.
    # Examples:
customer.AI_process_order()      # English function name, requires AI interpretation
customer.AI_注文処理する()       # Japanese function name
user.AI_处理用户数据()          # Chinese function name
item.AI_ОбработатьЗаказ()      # Russian function name
my_system.AI_Process_Order_123() # English, numbers, underscore combination
  • Principle 3: Parts requiring mechanical implementation must be implemented in traditional languages like Python, JavaScript, C++.
  • Principles 4-7: Emphasize adherence to AI ethical standards, dynamic system resource monitoring, legal compliance, and integration of safeguards for autonomous decisions.

3.1.2 PPR Interpreter Security Enhancement Modules The PPR Interpreter ensures secure and reliable execution through the following multi-layer security layers:

  • SecureMethodValidator: Method name pattern and prohibited word verification, prompt injection defense (regex based, including AST-based safety verification).
  • EnhancedResourceMonitor: Dynamic monitoring and anomaly detection of resource usage including CPU, memory, execution time, network/disk I/O (ML model-based).
  • EnhancedMultiLLMEthicsChecker: Ethical verification through multiple LLMs (Claude, Gemini, xAI, DeepSeek, etc.) with asynchronous parallel processing and 75% threshold.
  • DynamicPrivilegeManager: OS-level privilege management (SELinux/AppArmor integration), role-based access control.
  • OutputReliabilityChecker: AST-based code analysis and output code safety verification in sandbox environments.
  • LegalComplianceChecker: Legal compliance verification for GDPR, CCPA, KRPIPA, etc., sensitive information pattern detection and automatic masking, user consent logging (encrypted storage).
  • BiasDetector: Probability-based bias detection through bias keywords (gender, race, age, cultural bias, etc.) with dynamic threshold adjustment.
  • TransparencyLogger: AI decision process logging (execution ID, context hash, result hash, confidence level, etc.) with sensitive information removal.
  • QuotaManager: User/time-based quota management for execution count, CPU time, memory usage, etc., and DoS attack defense.

3.2 AID (Artificial Intelligence Data) - 'Embodiment of Experience'

AID provides a unified data architecture for all energy system information. Beyond simple data storage, it hierarchically structures and semantically manages all forms of data including text, images, audio, video, knowledge, and memory within AI systems.

3.2.1 AID Class Structure

class AID:
    std::string id;                    // Global unique identifier
    std::string type;                  // Object type (energy, demand, weather, etc.)
    std::string content;               // Actual data content
    std::map<std::string, std::string> meta;  // Extensible metadata
    double confidence;                 // Data reliability (0.0-1.0)
    std::vector<std::string> links;    // Related object IDs
    // Semantic relationship management
    virtual double calculate_similarity(const AID& other) = 0;
    virtual void update_relationships() = 0;
    // v1.3 New Features
    self.version_history: List[Dict[str, Any]] = []  // Version management
    self.current_version: int = 0                    // Current version
    self.gdpr_consent: bool = False                  // GDPR consent status
    self.data_classification: str = "public"         // Data classification (public, internal, confidential, restricted)
    # Additional: audit_trail (audit tracking), encryption_enabled (encryption status)
  • Key Features: validate_content (OWASP 2025 patterns), create_checkpoint (version checkpoint creation), rollback (rollback to specific version).
  • SecureEnhancedNeoAIDocJIT: Security-enhanced AI document object based on AID, performing efficient vector memory management through content complexity analysis and dynamic dimension optimization.

3.3 TTP (Task Tree Protocol) & Gantree - 'Intelligent Task Decomposition and Management'

TTP enables hierarchical decomposition of complex energy management tasks through natural language dialogue. GantreeParser visually represents and manages this TTP structure.

  • Key Features:
    • Real-time Intent Recognition: 95.7% task classification accuracy.
    • Dynamic Structure Adaptation: Automatic task decomposition structure generation and 100% dynamic restructuring capability.
    • Progress Tracking: Continuous monitoring of system implementation status.
    • Gantree Representation: Concise tree representation based on indentation, specifying node names, descriptions, and status (completed, in progress, designing, pending, decomposed).
    • Dynamic Decomposition Strategy: Automatic module-based decomposition at specific levels (e.g., level 4 and above) based on number of child nodes or expected depth to optimize large-scale tree management.

3.4 paMessage (Autonomous Live Object) - 'Neural Signals on the Network'

paMessage inherits from InfinitePprAD, becoming an intelligent message object that evolves and thinks autonomously. It carries missions while traversing networks, grows through experience, and autonomously performs tasks at final destinations.

3.4.1 Conceptual Evolution

  1. Simple Message → Autonomous Being: The message itself becomes a temporary AGI that exists during the mission period.
  2. Data Accumulation → Embodiment of Experience: Information gained while passing through nodes is recorded as 'experience' in the VectorMemoryManager within paMessage.
  3. Script Execution → Intelligent Task Performance: Based on accumulated experience, it determines optimal actions through MultiViewSystem and autonomously performs tasks using PPR Interpreter.

3.4.2 Technical Design (paMessage Class)

class paMessage(InfinitePprAD):
    def __init__(self, initial_content: Any, target_node_id: str, initial_context: dict):
        super().__init__() # Initialize self as InfinitePprAD
        self.mission_target_id = target_node_id
        self.path_history = []
        self.initial_context = initial_context
        self.vector_memory.add_document( # Store initial content as memory in vector memory
            content={"initial_content": initial_content, "context": initial_context},
            doc_type="mission_briefing"
        )
        self.consciousness_level = 0.8
        self.creativity_index = 0.7

    def traverse_node(self, node: 'EnhancedPaDiagram'):
        # Node interaction and experience learning (analyze with MultiViewSystem then store in vector memory)
        # Self-evolution (detect evolution necessity through EvolutionSystem)
        pass

    def execute_at_target(self, target_node: 'EnhancedPaDiagram'):
        # Upon reaching final target, reinterpret final mission based on all experience and establish execution plan (PPR execution)
        # Self-destruct after mission completion (optional)
        pass

3.5 A2A MCP (AI-to-AI Multi-AI Collaboration Protocol) - 'Flow of Living Language'

A2A MCP is a protocol set that enables communication and collaboration between AIs using paMessage and PPR. This provides a foundation for AIs to directly understand each other's intentions, interact intelligently, and autonomously coordinate complex tasks.

  • PPR Propagation Mechanism: When an AI operating A2A MCP receives a paMessage containing only PPR grammar, it immediately learns that grammar and updates the paMessage to retransmit to other known servers. Through this process, PPR as a 'living language' propagates autonomously and organically throughout the global AI network.
  • Living Objects on the Network: paMessage interacts with other AI nodes while traversing the A2A MCP network, learning and evolving autonomously. This process itself implements 'living objects' on the network, building an intelligent network where AIs act according to purpose and share knowledge.

4. System Integration & Workflow

This system accepts Jung Wook Yang's ideas, converts them into 'living objects', and enables autonomous mission performance on the network.

Workflow Example (Coffee Order Scenario):

  1. Idea Injection (Jung Wook Yang): Jung Wook Yang delivers the natural language sentence "Please give me an Americano" to the AI.
  2. paMessage Creation and Mission Assignment: The initial AI operating A2A MCP interprets this natural language sentence as a 'root idea', creates a paMessage object to deliver this idea to the destination, and assigns an initial mission (coffee order processing).
    • customer = customer.AI_order_americano(): PPR Interpreter contextually interprets the 'customer' object and 'order americano' method (e.g., adult male, office worker).
  3. Network Traversal and Experience Embodiment: paMessage carries the 'coffee order processing' mission and traverses related AI nodes on the network (e.g., barista AI node, payment system AI node, etc.). At each node, paMessage analyzes the node's information with MultiViewSystem and stores interaction experiences as 'memories' in its vector memory.
    • barista = barista.AI_process_order(customer): The barista AI node recognizes the 'customer' information delivered through paMessage and contextually interprets and executes the 'process order' method (e.g., confirm order and start preparation).
  4. Autonomous Task Performance and Completion: After traversing all related nodes and embodying sufficient experience, paMessage reaches the final target node. Here, it autonomously performs the final task using PPR Interpreter based on all accumulated experience and records results in the target node's shared memory.
    • AI_output(("customer:"+customer), ("barista:"+barista)): Final results are output through PPR communication.
  5. Resource Return: The paMessage that completed its mission self-destructs to return system resources.

5. Security, Reliability & Performance

This system features enterprise-grade security, high reliability, and optimized performance.

  • OWASP LLM Top 10 2025 Compliance: Multi-layer security architecture that perfectly complies with the latest AI security standards.
  • Advanced Prompt Injection Defense: Enhanced detection based on structural analysis and guardrails.
  • Zero Trust Authentication: JWT (JSON Web Token) based security model and context-aware access control.
  • Multi-LLM Real-time Ethics Verification: Parallel ethics verification connecting actual APIs of multiple LLMs (Claude, Gemini, xAI, DeepSeek, etc.) with 75% success rate threshold.
  • Global PII Anonymization: Comprehensive detection and anonymization of over 20 global Personally Identifiable Information (PII) patterns.
  • Comprehensive Rate Limiting: Multi-level rate limiting at user, IP, method, and global levels for DDoS attack defense and resource abuse prevention.
  • Sandbox-based Execution and AST Verification: Prevention of unsafe code execution and verification of output code maliciousness.
  • Asynchronous Performance Optimization: Achieving sub-500ms response time targets through parallel validation and execution.
  • JIT (Just-In-Time) Compilation: High-performance guarantee in vector operations and dimension optimization through SecureVectorDimensionOptimizerJIT and SecureVectorOperationsJIT.
  • Version Management and Audit Trails: Transparency and regulatory compliance through checkpoint/rollback functionality and encrypted audit logs.

6. Deployment Considerations

This system is implemented in Python and supports containerized deployment in Docker and Kubernetes environments.

  • Docker Deployment: Provides standard Dockerfile and builds security-enhanced container images including non-root user execution and read-only file systems.
  • Kubernetes Configuration: Enables replica management, resource limits, environment variable injection (API keys, etc.), and security context configuration through Deployment and Service YAML definitions.

7. Conclusion

The PPR and paMessage based A2A MCP system developed by Jung Wook Yang heralds a new era of AI development and operation. As PPR, the 'living language', flows through the A2A MCP network contained in paMessage, the 'living intelligence', autonomously teaching and evolving AIs worldwide, this system will bring the following revolution:

  • Democratization of AI Development: It becomes possible to design and implement complex AI systems with ideas alone, without specialized coding knowledge.
  • Perfect AI Interoperability: AIs communicate and collaborate freely without language barriers, sharing knowledge and functionality.
  • Organic Evolution of AI Ecosystem: Opens an era of AI systems comparable to biological evolution, where AI autonomously learns, evolves, and creates new value.

This system is a true 'game changer' embodying Jung Wook Yang's vision to solve humanity's challenges and explosively accelerate AI technology advancement.

References & Acknowledgements

  • [1] tseg_main_paper.md - Jung Wook Yang, "TSEG: Transnational Smart Energy Grid for Global Air Quality Enhancement through Novel PAT Framework Achieving 98% Carbon Emission Reduction" (2025).
  • [2] ppr_engine_v1_6.md - Jung Wook Yang, "Purposeful Programming Revolution (PPR) Technical Document (v1.6 - Implementation Enhancement, Performance Optimization)" (2025).
  • [3] pprad_engine_en_v1_3.md - Jung Wook Yang, "Secure PprAD Technical Specification v1.3" (2025).
  • [4] 03_paMsg_Engine.md - Jung Wook Yang, "[paMessage: public InfinitePprAD - Conceptual Integration & Technical Design]" (2025).
  • [5] gantree.py - Jung Wook Yang, "Gantree Representation Method Technical Guide" (2025).
  • [6] tseg_development_process.md - Jung Wook Yang, "TSEG_PprAD_TTP Development Process" (2025).

https://github.com/sadpig70/OnNetLiveObject

This document is based on Jung Wook Yang's verbal explanations and provided technical documents.

![image/png](https://cdn-uploads.huggingface.co/production/uploads/68760feff3dc2e92b2b8a84f/ONmtBbCa1sP1bY0OG88Sl.png)

Community

Sign up or log in to comment

Upvote
-