home / skills / openclaw / skills / smart-memory-manager

smart-memory-manager skill

safe

This skill intelligently manages short and long-term memories for agents, enabling fast retrieval, automatic summaries, and reliable persistence.

npx playbooks add skill openclaw/skills --skill smart-memory-manager

Review the files below or copy the command above to add this skill to your agents.

Files (4)

SKILL.md

2.8 KB

---
name: smart-memory-manager
slug: smart-memory-manager
description: Intelligent memory management for agents with short/long-term memory layering, semantic search, auto summarization, RAG enhancement
---

# 🧠 智能记忆管理器
## 核心亮点
1. 📚 **分层记忆体系**：短期/长期/重要记忆三层架构，自动清理过期记忆，解决上下文溢出问题
2. 🔍 **多模式检索**：支持关键词/语义/混合三种检索模式，快速召回相关记忆，提升RAG准确率
3. 📝 **自动摘要能力**：一键生成记忆摘要，支持长会话上下文压缩，token占用减少70%
4. 💾 **持久化支持**：支持内存/磁盘持久化，重启后记忆不丢失

## 🎯 适用场景
- 长会话Agent、聊天机器人
- RAG应用的记忆层
- 需要长期记忆的任务型Agent
- 客服、助理类Agent的上下文管理

## 📝 参数说明
| 参数 | 类型 | 必填 | 说明 |
|------|------|------|------|
| action | string | 是 | 操作类型：add/search/summarize/clear/list/load/save |
| content | string | 否 | add操作必填，记忆内容 |
| type | string | 否 | add操作可选，记忆类型：short-term/long-term/important，默认short-term |
| query | string | 否 | search操作必填，搜索关键词 |
| limit | number | 否 | search/list操作可选，返回结果数量，默认5/20 |
| typeFilter | string | 否 | 所有操作可选，过滤记忆类型，默认all |
| persist | boolean | 否 | add操作可选，是否持久化存储，默认false |
| persistPath | string | 否 | load/save操作可选，持久化文件路径，默认./memory-store.json |

## 💡 开箱即用示例
### 添加记忆
```typescript
// 添加长期记忆
await skills.smartMemoryManager({
  action: "add",
  content: "用户喜欢喝咖啡，不加糖，每周三下午喝奶茶",
  type: "long-term",
  persist: true
});
```

### 搜索记忆
```typescript
const result = await skills.smartMemoryManager({
  action: "search",
  query: "用户喜好",
  limit: 3,
  searchMode: "hybrid" // 关键词+语义混合检索
});
```

### 生成会话摘要
```typescript
const summary = await skills.smartMemoryManager({
  action: "summarize",
  typeFilter: "short-term",
  maxTokens: 500
});
```

### 持久化与加载
```typescript
// 保存所有记忆到磁盘
await skills.smartMemoryManager({
  action: "save",
  persistPath: "./my-memory.json"
});

// 从磁盘加载记忆
await skills.smartMemoryManager({
  action: "load",
  persistPath: "./my-memory.json"
});
```

## 🔧 技术实现说明
- 内置记忆自动清理机制，短期记忆最多保留100条，避免内存溢出
- 模块化设计，可轻松对接向量数据库实现语义检索
- 全链路类型安全，参数自动校验
- 轻量无外部依赖，开箱即用，也支持自定义扩展

Overview

This skill provides an intelligent memory manager for agents, implementing layered short-term, long-term, and important memory with persistence and automatic cleanup. It boosts retrieval-augmented generation (RAG) accuracy through hybrid keyword and semantic search, plus on-demand summarization to compress long contexts and reduce token usage.

How this skill works

The manager stores memories in three tiers and enforces retention policies (e.g., short-term caps) to avoid context overflow. It supports keyword, semantic, and hybrid search modes, automatic summarization to compress conversations, and optional persistence to disk or memory stores. Parameters control actions like add/search/summarize/save/load and filters for type, limit, and persistence behavior.

When to use it

Agents that handle long-running or multi-session conversations and need persistent context.
RAG systems that require a reliable memory layer to improve retrieval relevance.
Task agents that must remember user preferences or important facts over time.
Customer support or assistant bots that need concise context summaries to stay within token budgets.

Best practices

Classify memories on add using short-term, long-term, or important to get correct retention behavior.
Use hybrid search for best recall when semantic matches and keywords both matter.
Periodically run summarize on short-term memories to compress sessions and save tokens.
Persist important memories to disk to survive restarts and for safe backups.
Tune search limits and retention caps based on agent memory footprint and expected session length.

Example use cases

Store and recall user preferences (e.g., dietary habits, likes/dislikes) for personalized responses.
Compress and summarize long chat sessions before passing context to a large model to reduce token cost.
Provide a memory layer for RAG pipelines to retrieve relevant past interactions during answer generation.
Save critical facts to disk so the agent retains them after redeployments or restarts.
Filter and list recent short-term memory to reconstruct conversation state for follow-up actions.

FAQ

How are memories cleaned up?

Short-term memories have a configured cap and older entries are pruned automatically; important and long-term memory persist until explicitly cleared or saved.

Can I use a vector DB for semantic search?

Yes. The design is modular and supports integration with external vector stores for scalable semantic retrieval.