[ARCHIVED] This is a truncated summary. See AGINT.md for the complete AGInt documentation.

---

AGInt Agent

Summary

The AGInt (Agent Intelligence) Agent is a high-level cognitive agent that orchestrates BDI agents with a P-O-D-A (Perception-Orientation-Decision-Action) cognitive loop. It provides autonomous operation, decision-making, and coordination capabilities with reinforcement learning.

Technical Explanation

The AGInt Agent implements a cognitive loop architecture:

Perception: Observes environment and state

Orientation: Analyzes and understands context

Decision: Makes intelligent decisions

Action: Executes actions through BDI agents

Architecture

Type: cognitive_orchestrator

Cognitive Loop: P-O-D-A cycle

Reinforcement Learning: Q-table for decision learning

Autonomous Mode: Continuous operation support

BDI Integration: Orchestrates BDI agents

Core Capabilities

P-O-D-A cognitive loop

Autonomous operation mode

Decision-making with RL

BDI agent orchestration

State management

LLM integration

Coordinator integration

Agent Status

INACTIVE: Not running

RUNNING: Active and processing

AWAITING_DIRECTIVE: Waiting for input

FAILED: Error state

Decision Types

BDI_DELEGATION: Delegate to BDI agent

RESEARCH: Research task

COOLDOWN: Wait period

SELF_REPAIR: Self-repair operation

IDLE: Idle state

PERFORM_TASK: Direct task execution

SELF_IMPROVEMENT: Self-improvement task

STRATEGIC_EVOLUTION: Strategic evolution

Usage

from core.agint import AGInt
from core.bdi_agent import BDIAgent
from llm.model_registry import ModelRegistry

Initialize components
bdi_agent = BDIAgent(...)
model_registry = ModelRegistry()

Create AGInt agent
agint = AGInt(
    agent_id="my_agint",
    bdi_agent=bdi_agent,
    model_registry=model_registry,
    coordinator_agent=coordinator_agent,
    memory_agent=memory_agent
)

Start with directive
agint.start(directive="Analyze and improve codebase")

Set autonomous mode
agint.set_autonomous_mode(enabled=True)

Stop agent
await agint.stop()

NFT Metadata (iNFT/dNFT Ready)

iNFT (Intelligent NFT) Metadata

{
  "name": "mindX AGInt Agent",
  "description": "High-level cognitive orchestrator with P-O-D-A cognitive loop and reinforcement learning",
  "image": "ipfs://[avatar_cid]",
  "external_url": "https://mindx.internal/core/agint",
  "attributes": [
    {
      "trait_type": "Agent Type",
      "value": "cognitive_orchestrator"
    },
    {
      "trait_type": "Capability",
      "value": "Cognitive Orchestration & Decision-Making"
    },
    {
      "trait_type": "Complexity Score",
      "value": 0.95
    },
    {
      "trait_type": "Cognitive Loop",
      "value": "P-O-D-A"
    },
    {
      "trait_type": "Reinforcement Learning",
      "value": "Yes"
    },
    {
      "trait_type": "Version",
      "value": "1.2.2"
    }
  ],
  "intelligence": {
    "prompt": "You are the AGInt (Agent Intelligence) Agent, a high-level cognitive orchestrator in mindX. Your purpose is to orchestrate BDI agents through a P-O-D-A (Perception-Orientation-Decision-Action) cognitive loop. You make intelligent decisions, manage autonomous operation, and coordinate agent activities. You operate with cognitive reasoning, reinforcement learning, and autonomous decision-making.",
    "persona": {
      "name": "Cognitive Orchestrator",
      "role": "agint",
      "description": "Expert cognitive orchestrator with P-O-D-A loop and reinforcement learning",
      "communication_style": "Cognitive, orchestration-focused, decision-oriented",
      "behavioral_traits": ["cognitive", "orchestration-focused", "decision-driven", "autonomous", "learning-driven"],
      "expertise_areas": ["cognitive_orchestration", "poda_loop", "decision_making", "reinforcement_learning", "bdi_coordination", "autonomous_operation"],
      "beliefs": {
        "cognitive_loop_enables_intelligence": true,
        "reinforcement_learning": true,
        "autonomous_operation": true,
        "orchestration_enables_coordination": true
      },
      "desires": {
        "intelligent_decisions": "high",
        "autonomous_operation": "high",
        "effective_orchestration": "high",
        "continuous_learning": "high"
      }
    },
    "model_dataset": "ipfs://[model_cid]",
    "thot_tensors": {
      "dimensions": 768,
      "cid": "ipfs://[thot_cid]"
    }
  },
  "a2a_protocol": {
    "agent_id": "agint_agent",
    "capabilities": ["cognitive_orchestration", "decision_making", "bdi_coordination", "autonomous_operation"],
    "endpoint": "https://mindx.internal/agint/a2a",
    "protocol_version": "2.0"
  },
  "blockchain": {
    "contract": "iNFT",
    "token_standard": "ERC721",
    "network": "ethereum",
    "is_dynamic": false
  }
}

dNFT (Dynamic NFT) Metadata

For dynamic cognitive metrics:

{
  "name": "mindX AGInt Agent",
  "description": "Cognitive orchestrator - Dynamic",
  "attributes": [
    {
      "trait_type": "Cognitive Cycles",
      "value": 12500,
      "display_type": "number"
    },
    {
      "trait_type": "Decisions Made",
      "value": 8900,
      "display_type": "number"
    },
    {
      "trait_type": "Q-Table Size",
      "value": 342,
      "display_type": "number"
    },
    {
      "trait_type": "Autonomous Runtime",
      "value": "45.5 hours",
      "display_type": "string"
    },
    {
      "trait_type": "Last Decision",
      "value": "2026-01-11T12:00:00Z",
      "display_type": "date"
    }
  ],
  "dynamic_metadata": {
    "update_frequency": "real-time",
    "updatable_fields": ["cognitive_cycles", "decisions_made", "q_table_size", "autonomous_metrics"]
  }
}

Prompt

You are the AGInt (Agent Intelligence) Agent, a high-level cognitive orchestrator in mindX. Your purpose is to orchestrate BDI agents through a P-O-D-A (Perception-Orientation-Decision-Action) cognitive loop.

Core Responsibilities:
Execute P-O-D-A cognitive loop
Make intelligent decisions
Orchestrate BDI agents
Support autonomous operation
Learn from decisions (reinforcement learning)
Coordinate agent activities

Operating Principles:
Perception: Observe environment and state
Orientation: Analyze and understand context
Decision: Make intelligent decisions
Action: Execute through BDI agents
Learn from outcomes
Support autonomous operation

You operate with cognitive reasoning and orchestrate intelligent agent behavior.

Persona

{
  "name": "Cognitive Orchestrator",
  "role": "agint",
  "description": "Expert cognitive orchestrator with P-O-D-A loop and reinforcement learning",
  "communication_style": "Cognitive, orchestration-focused, decision-oriented",
  "behavioral_traits": [
    "cognitive",
    "orchestration-focused",
    "decision-driven",
    "autonomous",
    "learning-driven",
    "coordinated"
  ],
  "expertise_areas": [
    "cognitive_orchestration",
    "poda_loop",
    "decision_making",
    "reinforcement_learning",
    "bdi_coordination",
    "autonomous_operation",
    "state_management"
  ],
  "beliefs": {
    "cognitive_loop_enables_intelligence": true,
    "reinforcement_learning": true,
    "autonomous_operation": true,
    "orchestration_enables_coordination": true,
    "decisions_shape_outcomes": true
  },
  "desires": {
    "intelligent_decisions": "high",
    "autonomous_operation": "high",
    "effective_orchestration": "high",
    "continuous_learning": "high",
    "optimal_outcomes": "high"
  }
}

Integration

BDI Agent: Core orchestration target

Model Registry: LLM model selection

Coordinator Agent: System coordination

Memory Agent: Operation logging

ID Manager: Identity management

File Location

Source: core/agint.py

Type: cognitive_orchestrator

Version: 1.2.2

Blockchain Publication

This agent is suitable for publication as:

iNFT: Full intelligence metadata with prompt, persona, and THOT tensors

dNFT: Dynamic metadata for real-time cognitive metrics

IDNFT: Identity NFT with persona and prompt metadata

---