GEO-INFER-AGENT

Instructions

Core Capabilities

• Agent base: Transport pattern for agent communication
• Active Inference: Agent-level free energy minimization
• Telemetry: JSON snapshot + per-agent metrics tracking
• Coordination: Message passing, shared beliefs, swarm protocols
• Rule-based: Decision-tree agents for simple spatial tasks

Key Imports

from geo_infer_agent.core.agent_base import GeoAgent
from geo_infer_agent.core.active_inference import ActiveInferenceAgent
from geo_infer_agent.core.telemetry import AgentTelemetry
from geo_infer_agent.models.rule_based import RuleBasedAgent

Examples

from geo_infer_agent.core.agent_base import GeoAgent
from geo_infer_agent.core.telemetry import AgentTelemetry

agent = GeoAgent(agent_id="explorer_01", position=(45.5, -122.6))
telemetry = AgentTelemetry(agent)

# Perception-action loop
observation = agent.perceive(environment_state)
action = agent.decide(observation)
agent.act(action)

# Snapshot telemetry as JSON
snapshot = telemetry.snapshot()
print(f"Steps: {snapshot['total_steps']}, Position: {snapshot['position']}")

from geo_infer_agent.core.active_inference import ActiveInferenceAgent
import numpy as np

ai_agent = ActiveInferenceAgent(n_states=8, n_observations=5, n_actions=4)
obs = np.random.dirichlet(np.ones(5))
action = ai_agent.act(obs)
print(f"Selected action: {action}, Free energy: {ai_agent.free_energy:.4f}")

Guidelines

• Telemetry uses JSON snapshots and per-agent metrics (not TODOs)
• Agent base uses transport pattern for communication
• Test: uv run python -m pytest GEO-INFER-AGENT/tests/ -v

Integrations

• ACT → Active Inference agent decision-making
• SIM → Multi-agent simulation environments
• ANT → Swarm intelligence coordination
• SPACE → Spatial state representation for agents
• APP → Agent configuration and control widgets
• OPS → Agent telemetry monitoring