🧠 GNOSTRA: The Sovereign Machine AGI

A decentralized, autonomous AGI system for optimal resource allocation to solve human needs

---

🚀 Quick Start (30 Seconds)

# Clone and navigate
cd project-gnosis

# Run basic allocation test
.\test-scenario1.ps1

# That's it! ✨

First run: 2-5 minutes (builds Docker containers)
Subsequent runs: 30-60 seconds

---

📋 What is GNOSTRA?

GNOSTRA is a Sovereign Machine - an autonomous AGI system designed to:

• Identify bottlenecks in human needs using a 12-pillar framework
• Optimally allocate resources using constraint programming (OR-Tools)
• Learn and adapt through reinforcement learning
• Verify all decisions with counterfactual analysis
• Provide cryptographic proof for every allocation

Core Philosophy

"Maximize the NeedIndex by resolving the primary bottleneck"

The system uses a non-compensatory objective function - it cannot ignore critical failures even if other areas are thriving. A 0% "Biological Viability" pillar cannot be offset by a 500% "Play" pillar.

---

🎯 Testing Scenarios

Scenario 1: Basic Allocation (M0)

What: Core allocation loop with cryptographic verification
Time: 2-5 min (first), 30-60 sec (subsequent)
Command:

.\test-scenario1.ps1

Validates:

• ✅ Resource allocation optimization
• ✅ Cryptographic signatures
• ✅ End-to-end data flow
• ✅ Neo4j storage
• ✅ NATS messaging

---

Scenario 2: M3 Learning Loop

What: Complete autonomous learning system
Time: 3-7 min (first), 1-2 min (subsequent)
Command:

.\test-scenario2-m3.ps1

Validates:

• ✅ Digital twin simulation
• ✅ Reinforcement learning (Multi-Armed Bandit)
• ✅ Policy adaptation
• ✅ Counterfactual evaluation
• ✅ Autonomous operation

---

🏗️ Architecture

┌─────────────────────────────────────────────────────────┐
│                    GNOSTRA AGI System                   │
└─────────────────────────────────────────────────────────┘

Infrastructure:
├── Neo4j 5.26 (Knowledge Graph)
└── NATS 2.10 (Message Bus)

Core Services:
├── API Host (Rust) - gRPC Server
├── Allocator (Python) - CP-SAT Optimization
└── Storage - Event log + Graph state

M3 Learning Loop:
├── Simulator - Digital twin scenarios
├── Learner - RL policy optimization
└── Evaluator - Counterfactual analysis

Human Interface:
└── Qt/QML HMI Client

---

📦 Requirements

For Docker Testing (Recommended)

• Docker Desktop
• PowerShell (Windows) or Bash (Linux/Mac)

For Local Development

• Rust 1.70+ & Cargo
• Python 3.11+
• Docker (for Neo4j & NATS)
• Qt 6.8+ (for HMI client)

---

📚 Documentation

Getting Started

• Docker Simplified Testing ⭐ START HERE
• Quick Test Guide - Alternative manual approach
• Testing Guide - Detailed scenarios

System Documentation

• Blueprint - Philosophy and architecture
• M3 Complete - Learning loop achievements
• Operators Manual - Production operations

Technical Reference

• Service Architecture
• API Reference
• Data Schemas - JSON Schema definitions

---

🛠️ Common Commands

# Run basic test
.\test-scenario1.ps1

# Run M3 learning loop
.\test-scenario2-m3.ps1

# Get allocator public key
.\get-allocator-public-key.ps1

# Stop all services
.\stop-all.ps1

# View logs
docker-compose logs -f allocator
docker-compose logs -f learner

# Check service status
docker-compose ps

# Clean up everything
.\stop-all.ps1  # Then select 'y' to remove volumes

---

🎓 Understanding the Output

M0 Tabletop Test Output

[PHASE 1: SETUP] - Services initialized
[PHASE 2: INGRESS] - Data validated and stored
[PHASE 3: NEEDS COMPILER] - Bottleneck identified
[PHASE 4: ALLOCATOR] - Optimal solution found
[PHASE 5: COMPLETE] - Plan ID: plan_abc123...

M3 Learner Output

[METRIC] Scenario started
[MAB Update] Q-value updated (learning)
[POLICY] New policy selected
[AUDIT] Regret calculated

Lower Q-values → Better policies
Lower regret → Better decisions

---

🔬 System Capabilities

Current (M3 Complete)

• ✅ Constraint-based optimization (CP-SAT)
• ✅ 12-pillar needs assessment
• ✅ Cryptographic verification (Ed25519)
• ✅ Reinforcement learning (MAB)
• ✅ Counterfactual evaluation
• ✅ Autonomous adaptation
• ✅ NATS messaging
• ✅ Neo4j storage
• ✅ gRPC API
• ✅ Qt HMI client

Roadmap

• ⏳ Privacy layer (PII vaults)
• ⏳ Zero-knowledge proofs
• ⏳ DID-based identity
• ⏳ Dispute resolution
• ⏳ Meta-governance

---

🤝 Contributing

This is an active research project. Contributions welcome!

Development Workflow

1. Make changes to code
2. Rebuild affected services:

   docker-compose build allocator
   docker-compose up -d allocator

3. Test changes:

   docker-compose logs -f allocator

---

📊 Project Status

Phase: M3 (Learning Loop) - ✅ Complete
Services: 9 microservices
Languages: Rust, Python
Infrastructure: Docker, Neo4j, NATS
Testing: Automated via Docker Compose

---

🎯 Success Metrics

Basic Test (Scenario 1):

• Allocation time: < 5 seconds
• Plan verification: < 500ms
• Success rate: 100%

M3 Loop (Scenario 2):

• Learning convergence: 50-100 iterations
• Regret at convergence: < 0.05
• System throughput: 10+ plans/minute

---

🔗 Quick Links

• Start Testing: .\test-scenario1.ps1
• View Logs: docker-compose logs -f
• Neo4j Browser: http://localhost:7474 (neo4j/password)
• NATS Monitor: http://localhost:8222
• Allocator API: http://localhost:8000/docs

---

📄 License

To be determined - Research project

---

🙏 Acknowledgments

Built with:

• OR-Tools - Constraint optimization
• Neo4j - Graph database
• NATS - Message broker
• Rust - Systems programming
• Python - Scientific computing
• Qt - GUI framework

---

Ready to test the AGI? Run .\test-scenario1.ps1 🚀

Last Updated: 2025-11-09

---

by ASTRA MATRIX