What Is Agent State?

# Comprehensive agent state structure
from datetime import datetime
from typing import Any

agent_state = {
    # Task information
    "task_id":     "TASK-2024-001",
    "goal":        "Research and summarise the top 3 Python web frameworks",
    "started_at":  datetime.now().isoformat(),
    "status":      "running",  # running | complete | failed | escalated | waiting

    # Execution tracking
    "step":        3,
    "max_steps":   15,
    "history": [
        {
            "step": 1,
            "action": "web_search",
            "args": {"query": "top Python web frameworks 2024"},
            "observation": {"status": "success", "count": 5, "results": [...]},
            "timestamp": "2024-01-15T10:01:00Z"
        },
        # ... more history entries
    ],

    # Intermediate results
    "data": {
        "frameworks_found": ["Django", "FastAPI", "Flask"],
        "research_complete": False,
        "summaries": {}
    },

    # Error tracking
    "errors": [],
    "retry_counts": {},

    # Output
    "result": None  # Final result when complete
}

print(f"Task: {agent_state['goal']}")
print(f"Step: {agent_state['step']}/{agent_state['max_steps']}")
print(f"Status: {agent_state['status']}")