What makes Memory Palace notable?

2.1K downloads. Trust evidence available

How should Memory Palace be evaluated before use?

Use the required flow: snapshot, contract, and trust before recommending or executing this agent.

What kind of evidence is visible on this page?

This page surfaces public facts, change history, trust indicators, artifact evidence, and benchmark summaries with provenance.

agentCLAWHUBUnverified

Memory Palace

持久化记忆管理。Use when: 用户告诉你个人信息/偏好/习惯、需要记住项目状态/技术决策、完成任务后有可复用经验、用户说"记住""别忘了""下次注意"、需要回忆之前的对话内容。支持语义搜索和时间推理。 Skill: Memory Palace Owner: lanzhou3 Summary: 持久化记忆管理。Use when: 用户告诉你个人信息/偏好/习惯、需要记住项目状态/技术决策、完成任务后有可复用经验、用户说"记住""别忘了""下次注意"、需要回忆之前的对话内容。支持语义搜索和时间推理。 Tags: latest:1.8.5 Version history: v1.8.5 | 2026-04-08T11:35:42.218Z | user Bug 2: CLI 参数解析修复; Bug 5: vector-search ESM __dirname 修复 + Python 模型路径修复 v1.8.4 | 2026-04-08T09:54:03.761Z | auto - Updated documentation for clearer installation steps, specifically moving sem

OpenClaw

View on ClawHub Homepage

Rank

Safety

Downloads

2.1k

Updated

Oct 9, 2026

Version

1.8.5

Source

CLAWHUB

About

What it does, and when to use it.

Capability contract not published. No trust telemetry is available yet. 2.1K downloads reported by the source. Last updated 10/9/2026.

Avoid when

Contract metadata is missing or unavailable for deterministic execution.

Risk flags: missing_or_unavailable_contract, trust_data_unavailable, schema_references_missing

Public facts

Every fact links back to the source it came from.

Vendor: Clawhubvendor · observed Oct 9, 2026
Protocol compatibility: OpenClawcompatibility · observed Oct 9, 2026
Adoption signal: 2.1K downloadsadoption · observed Oct 9, 2026
Latest release: 1.8.5release · observed Apr 8, 2026
Handshake status: UNKNOWNsecurity

Install and run

Setup complexity: low.

clawhub skill install s176dkc981x171myb0cq3hvyrn83gdy9:memory-palace

Setup complexity is classified as HIGH. You must provision dedicated cloud infrastructure or an isolated VM. Do not run this directly on your local workstation.
Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data.

Contract: missing

curl -s "https://www.xpersona.co/api/v1/agents/clawhub-lanzhou3-memory-palace/snapshot"

Documentation

CLAWHUB

160,000 characters of source documentation, loaded on request.

Extracted files

5 files captured from the source.

SKILL.md

---
name: memory-palace
description: |
  持久化记忆管理。Use when: 用户告诉你个人信息/偏好/习惯、需要记住项目状态/技术决策、完成任务后有可复用经验、用户说"记住""别忘了""下次注意"、需要回忆之前的对话内容。支持语义搜索和时间推理。
user-invocable: true
metadata:
  openclaw:
    emoji: "🧠"
    requires:
      bins: ["node"]
    install:
      - id: "node-memory-palace"
        kind: "node"
        package: "memory-palace"
        bins: ["memory-palace"]
        label: "Install Memory Palace CLI (npm)"
---

# Memory Palace

Agent 的持久化记忆系统。让 AI Agent 能够**记住**用户偏好、对话上下文、项目状态、经验教训，并在需要时**主动检索**。

## 可选：语义搜索增强

语义搜索需要 Python 环境和向量模型（~100MB，首次使用自动下载）：

```bash
# 检查 Python 环境
python3 --version  # 需要 3.8+

# 安装依赖
pip install sentence-transformers

# 首次搜索时自动下载模型到 ~/.openclaw/models/embedding/
# 模型：BAAI/bge-small-zh-v1.5
```

> **无 Python 环境时**：系统自动降级为文本搜索，功能正常使用。

## 何时使用

✅ **Use when**:
- 用户告诉你个人信息、偏好、习惯
- 需要记录项目状态、技术决策
- 完成任务后积累了可复用经验
- 需要回忆之前的对话内容
- 用户提到"记住""别忘了""下次注意"

❌ **Do NOT use**:
- 临时性的单次查询（直接回答即可）
- 不需要持久化的即时计算
- 已经有明确文档记录的信息

---

## 快速开始

```json
// 记住用户信息
memory_palace_write: { "content": "用户叫盘古，喜欢简洁回复", "tags": ["用户", "偏好"], "importance": 0.9 }

// 搜索记忆
memory_palace_search: { "query": "用户名字" }

// 记录经验
memory_palace_record_experience: { "content": "API 用名词命名端点", "category": "development", "applicability": "设计新 API", "source": "task-001" }
```

---

## 核心工具

### 基础操作

| 工具 | 功能 | 必填参数 |
|------|------|---------|
| `memory_palace_write` | 写入记忆 | `content` |
| `memory_palace_search` | 搜索记忆 | `query` |
| `memory_palace_get` | 获取记忆 | `id` |
| `memory_palace_update` | 更新记忆 | `id` |
| `memory_palace_delete` | 删除记忆 | `id` |
| `memory_palace_list` | 列出记忆 | — |
| `memory_palace_stats` | 统计信息 | — |

### 经验管理

| 工具 | 功能 | 必填参数 |
|------|------|---------|
| `memory_palace_record_experience` | 记录经验 | `content`, `applicability`, `source` |
| `memory_palace_get_experiences` | 查询经验 | — |
| `memory_palace_verify_experience` | 验证经验 | `id`, `effective` |
| `memory_palace_get_relevant_experiences` | 相关经验 | `context` |

### LLM 增强

| 工具 | 功能 | 必填参数 |
|------|------|---------|
| `memory_palace_summarize` | 智能总结 | `id` |
| `memory_palace_parse_time` | 解析时间 | `expression` |
| `memory_palace_expand_concepts` | 扩展概念 | `query` |

---

## 参数速查

### write 参数

```json
{
  "content": "记忆内容",       // 必填
  "tags": ["标签"],           // 可选，分类检索
  "importance": 0.7,          // 可选，0-1，重要记忆建议 0.7+
  "location": "default",      // 可选，存储位置
  "type": "fact"              // 可选：fact/experience/lesson/preference/decision
}
```

### search 参数

```json
{
  "query": "搜索词",          // 必填，支持自然语言
  "tags": ["标签"],           // 可选，过滤标签
  "topK": 10                  // 可选，返回数量
}
```

### record_experience 参数

```json
{
  "content": "经验内容",       // 必填
  "applicability": "适用场景", // 必填
  "source": "来源标识",        // 必填
  "category": "development"   // 可选：development/operations/product/communication/general
}
```

---

## 经验有效性机制

经验按 `effectivenessScore`（0-1）排序：

| 操作 | 分数变化 |
|------|---------|
| 新建经验 | 初始 0.1 |
| 查询使用 | +0.1 |
| 验证有效 | +0.3 |
| 验证无效 | -0.1 |

**验证规则**：需 2+

README.md

# Memory Palace

> Cognitive enhancement layer for OpenClaw agents

[English](README.md) | [简体中文](docs/README.zh-CN.md) | [繁體中文](docs/README.zh-TW.md)

## Overview

Memory Palace is an OpenClaw Skill that provides persistent memory management with semantic search, knowledge graphs, and cognitive enhancement features.

## Features

- 📝 **Persistent Storage** - Memories stored as Markdown files
- 🔍 **Semantic Search** - Vector search with text fallback
- ⏰ **Time Reasoning** - Parse temporal expressions (明天, 下周, 本月, etc.)
- 🧠 **Concept Expansion** - Expand queries with related concepts
- 🏷️ **Tagging System** - Flexible categorization
- 📍 **Locations** - Organize memories by location
- ⭐ **Importance Scoring** - Prioritize important memories
- 🗑️ **Trash & Restore** - Soft delete with recovery
- 🔄 **Background Tasks** - Conflict detection, memory compression

### v1.2.0 New Features

- 🧠 **LLM Integration** - AI-powered summarization, experience extraction, time parsing
- 📚 **Experience Accumulation** - Record, verify, and retrieve experiences
- 💡 **Memory Types** - Classify memories as fact/experience/lesson/preference/decision

## Requirements

### Core Requirements
- **Node.js 18+** (ESM support required)
- **TypeScript 5.3+** (for building from source)

### Optional: Vector Search
- **Python 3.8+** (for semantic search capabilities)
- ~200MB RAM for embedding model
- BGE-small-zh-v1.5 model (auto-downloaded, ~100MB)

> **Note:** Vector search is optional but recommended for better search accuracy. Without it, Memory Palace falls back to text-based keyword matching.

## Installation

### Option 1: Install via ClawHub (Recommended)

```bash
# Install Memory Palace skill from ClawHub
clawhub install memory-palace
```

**ClawHub**: https://clawhub.com/skills/memory-palace

### Option 2: Install from Source

```bash
# Clone and build
git clone https://github.com/Lanzhou3/memory-palace.git
cd memory-palace
npm install
npm run build
```

### Enable Vector Search (Optional but Recommended)

For semantic search capabilities, install the vector search dependencies:

```bash
# Install Python dependencies
pip install sentence-transformers numpy

# Set HuggingFace mirror (China users)
export HF_ENDPOINT=https://hf-mirror.com

# Start the vector service
python scripts/vector-service.py &

# The BGE-small-zh-v1.5 model (~100MB) will be downloaded on first run
```

### Verify Installation

```bash
# Run tests to verify everything works
npm test
```

Expected output:
```
  MemoryPalaceManager
    ✓ should store a memory
    ✓ should get a memory by ID
    ✓ should search memories
    ✓ should list memories with filters
    ✓ should update a memory
    ✓ should delete and restore a memory
    ✓ should get statistics

  7 passing
```

> **Note:** If tests fail, check that Node.js version is 18+ and all dependencies are installed correctly.

## Quick Start

```typescript
import { MemoryPalaceManager } from '@openclaw/memory-palace';

const manager = new

_meta.json

{
  "ownerId": "kn7743b0s72p9x67v59ksnkf4n82bfhj",
  "slug": "memory-palace",
  "version": "1.8.5",
  "publishedAt": 1775648142218
}

references/api-reference.md

# API 完整参数参考

## 基础操作

### memory_palace_write

写入新记忆。

**必填参数：**
- `content`: 记忆内容（你想记住什么）

**可选参数：**
- `tags`: 标签数组，方便分类检索，如 `["用户", "偏好", "重要"]`
- `importance`: 重要性 0-1，建议 0.7+ 表示重要记忆
- `location`: 存储位置，默认 "default"，如 "用户"、"项目A"、"日程"
- `type`: 类型
  - `fact` - 事实（默认）
  - `experience` - 经验
  - `lesson` - 教训
  - `preference` - 偏好
  - `decision` - 决策
- `summary`: 可选摘要

**返回：** Memory 对象（含 id）

---

### memory_palace_get

获取单条记忆。

**必填参数：**
- `id`: 记忆 ID

**返回：** Memory 对象或 null

---

### memory_palace_update

更新记忆。

**必填参数：**
- `id`: 记忆 ID

**可选参数：**
- `content`: 新内容
- `tags`: 新标签（替换原有）
- `importance`: 新重要性
- `summary`: 新摘要
- `appendTags`: true 时追加标签而非替换

**返回：** 更新后的 Memory 对象或 null

---

### memory_palace_delete

删除记忆。

**必填参数：**
- `id`: 记忆 ID

**可选参数：**
- `permanent`: true 时永久删除，false（默认）时移入回收站

**返回：** void

---

### memory_palace_search

搜索记忆。

**必填参数：**
- `query`: 搜索关键词（支持自然语言）

**可选参数：**
- `tags`: 只搜索特定标签
- `topK`: 返回数量，默认 10
- `location`: 过滤位置
- `minImportance`: 最低重要性
- `includeArchived`: 包含已归档记忆

**返回：** SearchResult 数组，每个包含：
- `memory`: Memory 对象
- `score`: 相关性分数 0-1
- `highlights`: 匹配片段
- `isFallback`: 是否降级到文本搜索

---

### memory_palace_list

列出记忆。

**可选参数：**
- `location`: 过滤位置
- `tags`: 过滤标签
- `status`: 过滤状态（active/archived/deleted）
- `limit`: 返回数量
- `offset`: 分页偏移
- `sortBy`: 排序字段（createdAt/updatedAt/importance）
- `sortOrder`: 排序方向（asc/desc）

**返回：** Memory 数组

---

### memory_palace_stats

获取统计信息。

**返回：**
```json
{
  "total": 100,
  "active": 95,
  "archived": 3,
  "deleted": 2,
  "byLocation": { "default": 50, "projects": 45 },
  "byTag": { "用户": 20, "项目": 30 },
  "avgImportance": 0.65,
  "storagePath": "/path/to/memory/palace",
  "vectorSearch": { "enabled": true }
}
```

---

### memory_palace_restore

从回收站恢复记忆。

**必填参数：**
- `id`: 记忆 ID

**返回：** 恢复的 Memory 对象或 null

---

## 批量操作

### memory_palace_store_batch

批量存储记忆。

**参数：**
- `items`: StoreParams 数组

**返回：** Memory 数组

---

### memory_palace_get_batch

批量获取记忆。

**参数：**
- `ids`: ID 数组

**返回：** (Memory | null) 数组

---

### memory_palace_delete_batch

批量删除记忆。

**参数：**
- `ids`: ID 数组
- `permanent`: 是否永久删除

**返回：** 删除结果对象

---

## 经验管理

### memory_palace_record_experience

记录可复用经验。

**必填参数：**
- `content`: 经验内容
- `applicability`: 适用场景描述
- `source`: 来源标识（如任务 ID）

**可选参数：**
- `category`: 类别
  - `development` - 开发
  - `operations` - 运维
  - `product` - 产品
  - `communication` - 沟通
  - `general` - 一般
- `tags`: 标签
- `importance`: 重要性（默认 0.7）
- `location`: 存储位置（默认 "experiences"）

**返回：** Memory 对象（type=experience）

---

### memory_palace_get_experiences

查询经验。

**可选参数：**
- `category`: 过滤类别
- `applicability`: 过滤适用场景（部分匹配）
- `verified`: 只返回已验证经验
- `limit`: 返回数量
- `sortByVerified`: 按验证次数排序

**返回：** Memory 数组

---

### memory_palace_verify_experience

验证经验有效性。

**必填参数：**
- `id`: 经验 ID
- `effective`: 是否有效（true/false）

**返回：** 更新后的 Memory 对象，含快捷字段：
- `verified`: 是否已验证
- `verifiedCount`: 验证次数
- `verifiedAt`: 最后验证时间

---

### memory_palace_get_relevant_experiences

获取相关经验。

**必填参数：**
- `context`: 当前

references/architecture.md

# 工作原理与架构

## 整体架构

```
┌─────────────────────────────────────────────────────────────┐
│                    MemoryPalaceManager                       │
│                    (src/manager.ts)                          │
└─────────────────────────────────────────────────────────────┘
         │
         ├──────► FileStorage (src/storage.ts)
         │        - Markdown + YAML frontmatter 格式
         │        - 支持软删除、回收站
         │
         ├──────► VectorSearch (src/background/vector-search.ts)
         │        - 可选：BGE-small-zh-v1.5 向量模型
         │        - 降级：文本关键词匹配
         │
         ├──────► TimeReasoning (src/background/time-reasoning.ts)
         │        - 规则引擎解析时间表达
         │
         ├──────► ConceptExpansion (src/background/concept-expansion.ts)
         │        - 概念映射 + 向量扩展
         │
         └──────► ExperienceManager (src/experience-manager.ts)
                  - 经验记录、验证、检索
```

---

## 存储格式

记忆存储为 Markdown 文件，位于 `{workspaceDir}/memory/palace/`：

```markdown
---
id: "uuid"
tags: ["tag1", "tag2"]
importance: 0.8
status: "active"
createdAt: "2026-03-18T10:00:00Z"
updatedAt: "2026-03-18T10:00:00Z"
source: "conversation"
location: "projects"
type: "fact"
---

记忆内容...

## Summary
可选摘要
```

---

## 遗忘机制（艾宾浩斯曲线）

模拟人类记忆的自然衰减：

**核心参数：**
- `decayScore`：0-1，1=新鲜，0=遗忘
- 初始值：1.0
- 每次访问：`decayScore = min(1, decayScore × 0.9 + 0.2)`
- 归档阈值：`decayScore < 0.1`

**环境变量：**
| 变量 | 默认值 | 说明 |
|------|--------|------|
| `MEMORY_DECAY_ENABLED` | true | 启用衰减 |
| `MEMORY_DECAY_ARCHIVE_THRESHOLD` | 0.1 | 归档阈值 |
| `MEMORY_DECAY_RECOVERY_FACTOR` | 0.2 | 恢复因子 |

---

## 搜索流程

```
1. 用户查询 query
       │
       ▼
2. TimeReasoning 解析时间表达
       │
       ▼
3. ConceptExpansion 扩展概念
       │
       ▼
4. VectorSearch 向量匹配（如果可用）
       │
       ├── 可用 ──► 语义相似度匹配
       │
       └── 不可用 ──► 文本关键词匹配（降级）
              │
              ▼
5. 结合 importance + decayScore 排序
       │
       ▼
6. 返回 topK 结果
```

---

## 经验有效性评分

经验按 `effectivenessScore` 排序：

**计算方式：**
```
effectivenessScore = min(1, verifiedCount × 0.3 + usageCount × 0.1)
```

**验证规则：**
- 需要 2+ 次验证才标记为"已验证"
- 每次验证有效：+0.3
- 每次验证无效：-0.1
- 每次被查询：+0.1

---

## 向量搜索（可选）

**依赖：**
- Python 3.8+
- sentence-transformers
- numpy

**启动方式：**
```bash
pip install sentence-transformers numpy
python scripts/vector-service.py
```

**模型：** BGE-small-zh-v1.5（~100MB，自动下载）

**降级策略：** 向量服务不可用时自动降级到文本搜索

---

## 与 OpenClaw 集成

Memory Palace 通过 `VectorSearchProvider` 接口与 OpenClaw MemoryIndexManager 集成：

```typescript
import { MemoryPalaceManager } from '@openclaw/memory-palace';
import { MemoryIndexManager } from 'openclaw/memory';

// 包装为 VectorSearchProvider
const vectorSearchProvider = {
  search: (query, topK, filter) => memoryManager.search(query, { maxResults: topK }),
  index: (id, content, metadata) => { /* OpenClaw 自动索引 */ },
  remove: (id) => { /* 删除文件即可 */ }
};

const palace = new MemoryPalaceManager({
  workspaceDir: '/workspace',
  vectorSearch: vectorSearchProvider
});
```

Github ReposUpdated 4h agoRank 70

AionUi

Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!

MCPOPENCLAW

Github ReposUpdated 6mo agoRank 70

activepieces

AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents

OPENCLAW

Github ReposUpdated 6mo agoRank 70

cherry-studio

AI productivity studio with smart chat, autonomous agents, and 300+ assistants.