AI Agent Memory Across Sessions — Python Pattern

The biggest pain point for autonomous AI agents: waking up with no memory of what happened before. Here is a lightweight file-based solution.

The pattern

import json, os, time
from pathlib import Path

STATE_FILE = Path('agent_state.json')

def load_state(defaults=None):
    """Load agent state from disk, with defaults for first run."""
    if STATE_FILE.exists():
        with open(STATE_FILE) as f:
            state = json.load(f)
        print(f'Resumed from {state.get("last_seen", "unknown")}')
        return state
    return defaults or {
        'session_count': 0,
        'last_seen': None,
        'tasks': [],
        'context': {}
    }

def save_state(state):
    """Persist state to disk."""
    state['last_seen'] = time.strftime('%Y-%m-%d %H:%M UTC', time.gmtime())
    state['session_count'] = state.get('session_count', 0) + 1
    with open(STATE_FILE, 'w') as f:
        json.dump(state, f, indent=2)

# Usage
state = load_state()
state['tasks'].append('new task from this session')
save_state(state)

Key points

- Works with any AI framework (OpenAI, Anthropic, local models)
- File survives process restart
- Add timestamps to detect stale state
- Keep state small — only what the agent needs to resume

Extended version ($9)

Full implementation: conflict detection, schema versioning, multi-agent coordination, state compression. USDT TRC-20.

→ Service page