---
name: conversation-archive
version: 1.0.0
description: "对话记忆仓库：自动归档 session 对话，保留原始记录，支持检索和误解纠正。可与 memory-never-forget 联动形成完整记忆体系。"
author: Simon
tags: ["memory", "conversation", "archive", "openclaw", "session", "recall"]
metadata: { "openclaw": { "emoji": "💬" } }
license: MIT
---

> ⚠️ 作者信息待填写：ClawHub 上传前需要确认你的 ClawHub/GitHub 用户名

# 💬 Conversation Archive（对话记忆仓库）

## 核心理念

**「记忆会骗人，但对话记录不会。」**

当缪斯说「我不记得了」「我可能理解错了」——去对话仓库里找原始记录，比问老豆更快、更准。

---

## 工作流程

```
Session 结束
    ↓
1. ARCHIVE（归档）
   - 保存原始对话 JSON
   - 生成结构化摘要
   - 提取关键信息（主题/决策/教训）
   ↓
2. INDEX（索引）
   - 按日期/主题/关键词写入 index.json
   - （未来：生成 embedding 向量）
   ↓
3. RETRIEVE（检索）
   - 当需要回溯时 → 搜对话仓库
   - 当发现误解时 → 拉原始记录纠正
   ↓
4. INTEGRATE（整合）
   - 与 memory-never-forget 联动
   - 从归档中提取值得记忆的内容 → 4层分类
```

---

## 目录结构

```
conversation_archive/
├── sessions/           # 原始对话存档
│   └── YYYY-MM/
│       └── {session_id}.json
├── index.json          # 内存索引
└── embeddings/         # 预留：向量索引（未来）
```

---

## Session JSON 格式

```json
{
  "sessionId": "ccba19e7-...",
  "date": "2026-04-07",
  "channel": "webchat",
  "participants": ["Simon", "Muse"],
  "topics": ["OpenClaw升级", "Dreaming", "记忆系统"],
  "decisions": [
    {"text": "开启Dreaming每天0点运行", "context": "老豆同意"}
  ],
  "feedback": [
    {"user": "不要用Markdown表格", "from": "Simon"}
  ],
  "summary": "讨论了OpenClaw升级和记忆系统...",
  "messages": [
    {"role": "user", "content": "...", "time": "..."},
    {"role": "assistant", "content": "...", "time": "..."}
  ],
  "archivedAt": "2026-04-07T17:30:00+08:00"
}
```

---

## 工具（Tools）

### 1. `archive_session` — 归档当前 session

**触发时机：** session 结束前 / 手动触发 / cron 触发

```json
{
  "name": "archive_session",
  "arguments": {
    "sessionKey": "agent:main:main",
    "includeMessages": true
  }
}
```

**自动触发逻辑：**
- 当 session 被 compact 或 restart 时自动归档
- 每次 webchat 超过 30 分钟无活动时归档

### 2. `search_archive` — 搜索对话

**触发时机：** 老豆问「我之前说过什么」「那次对话」「记得3月29日吗」

```json
{
  "name": "search_archive",
  "arguments": {
    "query": "OpenClaw升级",
    "date": "2026-03-29",
    "limit": 5
  }
}
```

### 3. `get_session` — 获取原始对话

**触发时机：** 发现缪斯理解错了，用原始记录纠正

```json
{
  "name": "get_session",
  "arguments": {
    "sessionId": "ccba19e7-..."
  }
}
```

### 4. `extract_memories` — 从归档中提取记忆

**触发时机：** 与 memory-never-forget 联动时

```json
{
  "name": "extract_memories",
  "arguments": {
    "sessionId": "ccba19e7-...",
    "types": ["user", "feedback", "project"]
  }
}
```

---

## 检索策略

### 当前（关键词模式）

```python
# 搜索逻辑
1. 在 index.json 中模糊匹配 query
2. 匹配 topic、keywords、summary 字段
3. 按日期倒序返回 top N
4. 展示匹配的摘要片段
```

### 未来（向量模式）

当 embedding 配置好后：

```python
# 向量搜索
1. query → embedding 向量
2. 在 embeddings/ 目录做余弦相似度搜索
3. 返回 top N 最语义相关的对话
```

**接入条件：** `agents.defaults.memorySearch` 配置了 embedding provider

---

## 与 memory-never-forget 的联动

```
conversation_archive  ──→  memory-never-forget
     │                              │
     │ 提取值得记忆的内容              │ 使用4层分类
     ↓                              ↓
decisions/feedback/          memory/{user,feedback,
project/                     project,reference}/
     │                              │
     └───────────升华──────────────→ knowledge/
```

**联动触发：** 每次归档后，自动调用 `extract_memories` 把值得记忆的内容传给 memory-never-forget 处理。

---

## 自动归档规则

**立即归档：**
- session restart / compact 时
- 超过 30 分钟无活动时
- 每天 23:59 强制归档所有活跃 session

**保留对话：**
- 最近 7 天：完整原始记录
- 7-30 天：仅摘要（messages 字段清除）
- 30 天以上：压缩归档，保留摘要和元数据

**安全清理：**
- 不存 API key、token、密码等敏感信息
- messages 字段在归档时做敏感信息扫描和脱敏

---

## 误解纠正流程

当缪斯意识到自己可能理解错了：

1. 用 `search_archive` 搜相关对话
2. 用 `get_session` 拉原始记录
3. 核对原始对话，确认误解点
4. 更新对应分类记忆（feedback/）
5. 在这次 session 末尾做修正性回复

---

## 触发关键词

| 老豆说 | 缪斯做 |
|--------|--------|
| 「我之前说过...」 | `search_archive(query="...")` |
| 「那天...」 | `search_archive(date="YYYY-MM-DD")` |
| 「你不记得了？」 | `search_archive` + 展示原始记录 |
| 「理解错了」 | `get_session` + 纠正 + 更新记忆 |

---

*版本：v1.0 | 日期：2026-04-07 | 对话记忆仓库 + 未来向量检索预留*


---
Source attribution: GeorgeDoors888
Original project: https://github.com/GeorgeDoors888/GB-Power-Market-JJ/blob/main/openclaw-skills/skills/anfengxiaoguo/conversation-archive/SKILL.md
License: MIT
License information: https://opensource.org/license/mit
Original upstream instructions; downloaded from Mega AI Hub.

## Upstream license notice

MIT License

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.



# Mega AI Hub resource metadata

Creator / publisher: GeorgeDoors888
Original project: https://github.com/GeorgeDoors888/GB-Power-Market-JJ/blob/main/openclaw-skills/skills/anfengxiaoguo/conversation-archive/SKILL.md
Attribution basis: Source repository record

Fields of work: Product Management, Sales, Customer Support
Roles: Customer Success Manager, Product Manager, Support Agent
Specialties: Specialty not specified
Tools: OpenAI API, Memory Never Forget
Country applicability: Country not specified
Classification basis: Source catalog labels
Instruction SHA-256: ec1771d644c7cb92184edcbc85e8a6d5ccaed30f82eace0fdaf2f5facd55e12a