AutoDream-Core | 通用记忆整理引擎

让任何 Agent 都拥有记忆整理能力 | Give Any Agent Memory Consolidation Power

📖 概述 | Overview

中文: autodream-core 是一个通用的记忆整理引擎，采用适配器模式设计，支持跨平台记忆管理。 它解决了 AI 代理长期运行中的记忆衰减问题：

• 记忆文件随时间积累变得混乱
• 相对日期（如"昨天"）失去意义
• 过时的调试方案引用已删除的文件
• 矛盾条目堆积

English: autodream-core is a universal memory consolidation engine with adapter-based design for cross-platform memory management. It solves the memory decay problem in long-running AI agents:

• Memory files become chaotic over time
• Relative dates (e.g., "yesterday") lose meaning
• Outdated debug solutions reference deleted files
• Contradictory entries accumulate

🎯 核心特性 | Core Features

🚀 快速开始 | Quick Start

安装 | Installation

# 使用 skillhub (推荐)
skillhub install autodream-core

# 或使用 clawhub
clawhub install autodream-core

基础用法 | Basic Usage

from pathlib import Path
from autodream_core import AutoDreamEngine, OpenClawAdapter

# 初始化适配器 | Initialize adapter
adapter = OpenClawAdapter(
    workspace=Path("~/.openclaw/workspace-research").expanduser()
)

# 创建引擎 | Create engine
engine = AutoDreamEngine(adapter)

# 运行整理 | Run consolidation
result = engine.run(force=True)  # force=True 强制运行

print(f"整理完成！处理了 {result['consolidation']['final_count']} 个条目")

📦 目录结构 | Directory Structure

autodream-core/
├── SKILL.md           # 技能描述 (Skill metadata)
├── README.md          # 详细文档 (Documentation)
├── package.json       # 包信息 (Package info)
├── install.sh         # 安装脚本 (Install script)
├── core/              # 核心逻辑 (Core logic)
│   ├── engine.py      # 主引擎 (Main engine)
│   ├── analytics.py   # 分析日志 (Analytics)
│   ├── stages/        # 四阶段实现 (4 stages)
│   └── utils/         # 工具函数 (Utilities)
├── adapters/          # 平台适配器 (Platform adapters)
│   ├── base.py        # 基础接口 (Base interface)
│   └── openclaw.py    # OpenClaw 实现 (OpenClaw impl)
├── tests/             # 单元测试 (Unit tests)
└── examples/          # 使用示例 (Examples)

🔧 配置选项 | Configuration

from autodream_core import AutoDreamEngine, OpenClawAdapter

adapter = OpenClawAdapter(
    workspace=Path("/path/to/workspace"),
    memory_dir=Path("/path/to/workspace/memory"),  # 可选
)