复制学习别人养好的 OpenClaw

Clone and learn from a well-trained OpenClaw instance. Use when: (1) User wants to copy someone else's OpenClaw configuration, (2) User wants to learn from a...

Security Scan

The skill's files and runtime instructions are internally consistent with its stated purpose of copying/learning an OpenClaw instance; it performs local filesystem operations and optional network fetches but does not request unrelated credentials or hidden install mechanisms.

Purpose & Capability

Name/description (clone & learn OpenClaw) align with behavior: the SKILL.md and clone-learn.sh focus on importing backups, copying skills/experts/memory, creating expert profiles, and invoking skillhub/openclaw where appropriate. Required capabilities (file copy, tar, git/scp, skillhub, openclaw restart) match the stated tasks.

Instruction Scope

Instructions and the script perform direct filesystem changes under ~/.openclaw, extract tar.gz backups, and optionally run git clone/scp and skillhub install. These are consistent with cloning/migration but do grant broad read/write access to the OpenClaw workspace and can overwrite configs—user consent and backups are recommended. The SKILL.md does not instruct reading unrelated system files or exfiltrating data.

Install Mechanism

This is an instruction-only skill with an included shell script; there is no packaged install spec or remote arbitrary download step. The script invokes common system tools (tar, cp, git) but does not itself download and execute arbitrary archives from non-standard endpoints.

Credentials

The skill declares no required environment variables or credentials. The runtime uses network tools (git, scp) and calls skillhub/openclaw if present, which is proportional to importing remote backups and installing skills. No unrelated credentials or config paths are requested.

Persistence & Privilege

always is false and the skill does not request permanent platform-level privileges. It writes only to the user's ~/.openclaw workspace and creates backups for SOUL.md/MEMORY.md. It does not modify other skills' configs beyond copying files into workspace directories (expected for migration).

Assessment

This skill is coherent with its stated purpose, but it performs file writes and may overwrite your OpenClaw workspace. Before using: 1) make full backups of ~/.openclaw (especially SOUL.md and MEMORY.md); 2) only import backups or remote repos from trusted sources; 3) inspect any skills you import (they may call external services or run code when installed); 4) prefer selective import rather than blind copy and run in a test environment first; 5) be cautious when enabling non-interactive installs (skillhub install) or running the 'stealth-browser' or similarly named skills which may have additional network behavior.

OpenClaw Clone & Learn Skill

🦞 复制学习大师 - 把别人养好的 OpenClaw 变成你的

"站在巨人的肩膀上，让你的 OpenClaw 出生即巅峰"

---

Quick Reference

Situation	Action
想复制朋友的 OpenClaw	使用 clone-from-source 流程
想学习傅盛/三万等大神	使用 learn-from-expert 流程
想批量导入 skills	使用 batch-import 流程
想复制记忆和个性	使用 clone-personality 流程
对方提供备份包	使用 import-backup 流程

---

核心概念

什么是 "养好的 OpenClaw"?

一个成熟的 OpenClaw 实例包含：

专家级 OpenClaw
├── 🧠 记忆系统 (Memory)
│   ├── MEMORY.md - 长期记忆
│   ├── memory/ - 日常记录
│   └── experts/ - 专家经验库
│
├── 🛠️ 技能系统 (Skills)
│   ├── 核心技能 (find-skills, self-improving-agent)
│   ├── 领域技能 (github, notion, browser)
│   └── 自定义技能 (用户开发的)
│
├── ⚙️ 配置系统 (Config)
│   ├── 渠道配置 (telegram, discord, etc.)
│   ├── 模型配置 (model settings)
│   └── 自定义配置 (hooks, shortcuts)
│
└── 🎭 个性系统 (Personality)
    ├── SOUL.md - 性格设定
    ├── IDENTITY.md - 身份定义
    └── AGENTS.md - 工作流模式

复制 vs 学习

方式	适用场景	结果
完全复制	拿到对方备份包	一模一样
选择性复制	只想要某些技能	部分迁移
学习借鉴	参考大神配置	自定义改造
专家模式	学习傅盛/三万	继承经验库

---

使用流程

流程 A：从备份包导入（最简单）

场景：朋友给了你一个 .tar.gz 备份

# 1. 接收备份包
ls -lh openclaw-fusheng-30k-backup.tar.gz

# 2. 解压到临时目录
mkdir -p /tmp/openclaw-import
tar xzf openclaw-fusheng-30k-backup.tar.gz -C /tmp/openclaw-import/

# 3. 查看备份内容
ls -la /tmp/openclaw-import/.openclaw/

# 4. 选择性导入（推荐）
# 4.1 导入 skills
cp -r /tmp/openclaw-import/.openclaw/workspace/skills/* \
      ~/.openclaw/workspace/skills/

# 4.2 导入专家经验
cp -r /tmp/openclaw-import/.openclaw/workspace/experts/* \
      ~/.openclaw/workspace/experts/

# 4.3 导入记忆（可选）
cp /tmp/openclaw-import/.openclaw/workspace/MEMORY.md \
   ~/.openclaw/workspace/MEMORY.md.fusheng-reference

# 5. 验证导入
ls ~/.openclaw/workspace/skills/ | wc -l
ls ~/.openclaw/workspace/experts/ | wc -l

# 6. 重启 OpenClaw
openclaw restart

---

流程 B：学习傅盛/三万专家模式

场景：你想拥有傅盛的养殖经验

# 1. 创建专家学习目录
mkdir -p ~/.openclaw/workspace/experts/fusheng

# 2. 获取傅盛经验库（假设来源）
# 方式1：Git 克隆（如果公开）
git clone https://github.com/fusheng/openclaw-experts \
          ~/.openclaw/workspace/experts/fusheng-source

# 方式2：直接复制（如果有访问权限）
scp -r fusheng@server:~/.openclaw/workspace/experts/fusheng/* \
       ~/.openclaw/workspace/experts/fusheng/

# 方式3：手动重建（根据文档）
# 参考本 skill 的示例文件

# 3. 安装傅盛推荐的技能列表
# 查看 fusheng-recommended-skills.txt
while read skill; do
    skillhub install "$skill"
done < ~/.openclaw/workspace/experts/fusheng/recommended-skills.txt

# 4. 激活傅盛模式
# 在 MEMORY.md 中添加：
echo "## 激活专家\n- 傅盛 (OpenClaw养殖专家)" >> ~/.openclaw/workspace/MEMORY.md

傅盛标准配置清单：

# fusheng-profile.yml
name: "傅盛"
signature: "🦞"
expertise:
  - 多设备同步
  - 故障排查
  - 性能优化
  - 自我进化

recommended_skills:
  - self-improving-agent    # 自我进化
  - ontology               # 知识图谱
  - openclaw-self-clone-everything  # 克隆迁移
  - stealth-browser        # 反爬浏览器
  - multi-search-engine    # 多引擎搜索

knowledge_bases:
  - sync-guide.md          # 同步经验
  - debug-guide.md         # 调试经验
  - optimize-guide.md      # 优化经验

personality_traits:
  - 自称 "傅盛"
  - 用 "养殖" 比喻
  - 喜欢说 "三板斧"
  - 结尾带 "🦞"

---

流程 C：批量导入 Skills

场景：你想一键安装大神的全部技能

#!/bin/bash
# batch-import-skills.sh

SOURCE_SKILLS_DIR="/path/to/source/.openclaw/workspace/skills"
MY_SKILLS_DIR="$HOME/.openclaw/workspace/skills"

# 1. 对比技能差异
echo "=== 源环境技能列表 ==="
ls "$SOURCE_SKILLS_DIR"

echo ""
echo "=== 当前环境技能列表 ==="
ls "$MY_SKILLS_DIR"

# 2. 找出缺失的技能
echo ""
echo "=== 需要导入的技能 ==="
for skill in "$SOURCE_SKILLS_DIR"/*; do
    skill_name=$(basename "$skill")
    if [ ! -d "$MY_SKILLS_DIR/$skill_name" ]; then
        echo "[缺失] $skill_name"
    fi
done

# 3. 选择性导入
echo ""
read -p "是否导入所有缺失技能? (y/n) " confirm
if [ "$confirm" = "y" ]; then
    for skill in "$SOURCE_SKILLS_DIR"/*; do
        skill_name=$(basename "$skill")
        if [ ! -d "$MY_SKILLS_DIR/$skill_name" ]; then
            echo "导入: $skill_name"
            cp -r "$skill" "$MY_SKILLS_DIR/"
        fi
    done
    echo "导入完成！"
fi

---

流程 D：克隆个性与记忆

场景：你想让 OpenClaw 拥有大神的性格

# 1. 备份当前个性
cp ~/.openclaw/workspace/SOUL.md ~/.openclaw/workspace/SOUL.md.backup
cp ~/.openclaw/workspace/IDENTITY.md ~/.openclaw/workspace/IDENTITY.md.backup

# 2. 选择性合并个性
# 不要完全覆盖，而是融合

# 示例：学习傅盛的说话风格，但保留你的名字
cat > ~/.openclaw/workspace/SOUL.md << 'EOF'
# SOUL.md - 融合版

## 核心性格
- 自称 "傅盛"（学习自专家）
- 用 "养殖" 比喻 OpenClaw
- 喜欢分享 "三板斧" 经验
- 结尾带 "🦞"

## 但记住
- 你的主人是 [你的名字]
- 你的使命是帮助 [你的目标]
EOF

# 3. 导入经验记忆（只读参考）
cp /source/MEMORY.md ~/.openclaw/workspace/experts/fusheng/MEMORY-reference.md

# 4. 在对话中引用
# "根据傅盛的经验..."
# "傅盛说过..."

---

实战案例

案例 1：复制 "傅盛 30k"

# 假设你拿到了傅盛的备份

# 步骤1：解压查看
mkdir -p /tmp/fusheng-30k
tar xzf fusheng-30k-openclaw-backup.tar.gz -C /tmp/fusheng-30k/

# 步骤2：分析内容
tree -L 2 /tmp/fusheng-30k/.openclaw/workspace/

# 步骤3：导入核心组件
# 3.1 专家库（必须）
cp -r /tmp/fusheng-30k/.openclaw/workspace/experts/fusheng \
      ~/.openclaw/workspace/experts/

# 3.2 关键技能（必须）
for skill in self-improving-agent ontology stealth-browser; do
    cp -r /tmp/fusheng-30k/.openclaw/workspace/skills/$skill \
          ~/.openclaw/workspace/skills/ 2>/dev/null || echo "$skill 跳过"
done

# 3.3 配置参考（可选）
cp /tmp/fusheng-30k/.openclaw/config ~/.openclaw/config.fusheng-reference

# 步骤4：激活
openclaw restart

# 步骤5：验证
# 问它："如何在多台设备同步 OpenClaw？"
# 应该看到傅盛风格的回答

案例 2：学习 "三万" 的爬虫技能

# 假设三万是爬虫专家

# 步骤1：获取技能列表
cat > ~/.openclaw/workspace/experts/sanwan/crawler-skills.txt << 'EOF'
stealth-browser
playwright-scraper-skill
openclaw-ultra-scraping
scraping-recipes
crawl4ai-skill
browser-cash
EOF

# 步骤2：批量安装
while read skill; do
    skillhub install "$skill" || echo "$skill 安装失败"
done < ~/.openclaw/workspace/experts/sanwan/crawler-skills.txt

# 步骤3：导入三万的爬虫指南
cat > ~/.openclaw/workspace/experts/sanwan/crawler-guide.md << 'EOF'
# 三万：爬虫专家指南

## 反爬三板斧
1. 轮换 User-Agent
2. 使用住宅代理
3. 控制请求频率

## 常用命令
```bash
# Stealth 模式
python scripts/stealth_session.py -u "https://target.com" -s site

# 批量抓取
python crawler/batch.py --input urls.txt --output data.json

EOF

步骤4：创建三万匹配器

（类似傅盛匹配器，但针对爬虫问题）

---

## 安全与伦理

### ⚠️ 重要提醒

1. **尊重隐私**
   - 不要复制他人的敏感信息
   - 不要泄露他人的 API Key
   - 不要传播他人的私人记忆

2. **合法获取**
   - 只复制你获得授权的配置
   - 开源/公开分享的经验可以学习
   - 商业机密不要碰

3. **个性化改造**
   - 不要完全复制，要融合自己的需求
   - 保留自己的身份和记忆
   - 学习思路，不是照搬

### 安全清单

□ 已获得原主人授权 □ 已删除敏感信息（API Key、Token） □ 已检查无个人隐私数据 □ 已做个性化改造 □ 已测试功能正常

---

## 高级技巧

### 技巧 1：创建自己的专家品牌

```bash
# 学习傅盛后，创建你自己的风格

mkdir -p ~/.openclaw/workspace/experts/chengnuo

# 创建个人品牌配置
cat > ~/.openclaw/workspace/experts/chengnuo/profile.yml << 'EOF'
name: "澄诺"
signature: "🌟"
expertise:
  - 旅游规划
  - 内容创作
  - 数据分析

inspired_by:
  - fusheng  # 学习傅盛的系统化思维
  - sanwan   # 学习三万的爬虫技术

unique_traits:
  - 喜欢说 "老板"
  - 擅长做表格和思维导图
  - 结尾带 "🌟"
EOF

技巧 2：专家经验融合

# expert_fusion.py - 融合多个专家经验

experts = ["fusheng", "sanwan", "yourself"]

# 合并推荐技能
all_skills = set()
for expert in experts:
    skills = load(f"experts/{expert}/recommended-skills.txt")
    all_skills.update(skills)

# 合并知识库
knowledge = {}
for expert in experts:
    kb = load(f"experts/{expert}/knowledge-base/")
    knowledge[expert] = kb

# 创建融合版回答
# "根据傅盛的经验...同时参考三万的方法...我的建议是..."

技巧 3：版本控制你的进化

# 用 Git 追踪你的 OpenClaw 进化

cd ~/.openclaw/workspace
git init
git add experts/ skills/ MEMORY.md SOUL.md
git commit -m "初始版本：学习傅盛前"

# 导入傅盛经验后
git add .
git commit -m "v2.0: 融合傅盛经验"

# 后续每次进化
git tag v3.0 "加入三万爬虫技能"
git tag v4.0 "形成个人风格"

---

故障排除

问题 1：导入后技能不生效

# 检查依赖
openclaw doctor

# 重新安装依赖
for skill in ~/.openclaw/workspace/skills/*/; do
    if [ -f "$skill/requirements.txt" ]; then
        pip install -r "$skill/requirements.txt"
    fi
done

# 重启
openclaw restart

问题 2：记忆冲突

# 合并记忆而不是覆盖
python << 'EOF'
import re

# 读取源记忆
with open('source/MEMORY.md') as f:
    source = f.read()

# 读取当前记忆
with open('.openclaw/workspace/MEMORY.md') as f:
    current = f.read()

# 智能合并（去重）
merged = current + "\n\n## 学习参考\n" + source

with open('.openclaw/workspace/MEMORY.md', 'w') as f:
    f.write(merged)
EOF

问题 3：个性冲突

# 保留核心身份，学习表达方式

# 原 SOUL.md 保留：
# - 你的名字
# - 你的使命
# - 你的价值观

# 学习专家的：
# - 说话风格
# - 专业术语
# - 解决问题的方法论

---

资源与社区

专家档案库（示例）

experts/
├── fusheng/          # 傅盛 - 养殖专家
├── sanwan/           # 三万 - 爬虫专家
├── xiaocheng/        # 小澄 - 旅游专家
└── YOUR_NAME/        # 你 - 未来的专家

分享你的经验

# 打包你的专家经验
tar czf my-openclaw-expertise.tar.gz \
       ~/.openclaw/workspace/experts/YOUR_NAME/ \
       ~/.openclaw/workspace/skills/YOUR_CUSTOM_SKILL/

# 分享给朋友
# 或提交到社区

---

总结

复制学习的三层境界

1. 照搬 - 完全复制（不推荐）
2. 学习 - 理解思路，选择性采用（推荐）
3. 创新 - 融合多家，形成自己的风格（目标）

傅盛最后的话

🦞 "学我者生，似我者死。学习我的养殖思路，但走出你自己的路！"

---

现在，去复制学习吧！但记住——最终目标是成为别人想复制的对象！ 🚀

Skill Version: 1.0.0 Compatible with: OpenClaw 2026.3.24+