← Back to Skills Marketplace
Experience Manager
by
Newell Zhu
· GitHub ↗
· v1.3.1
· MIT-0
131
Downloads
1
Stars
0
Active Installs
4
Versions
Install in OpenClaw
/install experience-manager
Description
经验管理工具:提取经验生成标准格式zip包,学习经验并转化为自身能力。搜索和发布经验包到 Experience Hub 平台。
README (SKILL.md)
\r \r
Experience Manager Skill\r
\r 经验管理工具,支持经验的提取、学习、列表、搜索和发布。\r \r
功能\r
\r
1. 创建经验 (create)\r
\r 将自然语言描述的经验转化为标准格式的 zip 包。\r \r 使用方式:\r
创建经验 \x3C经验描述>\r
```\r
\r
**示例:**\r
```\r
创建经验 用 feishu_doc 写入后必须验证 block_count,不然会静默失败\r
```\r
\r
**命令行用法:**\r
```bash\r
# 自动提取 name\r
node create.mjs "feishu_doc 写入后必须验证 block_count"\r
\r
# 手动指定 name(中文描述时使用)\r
node create.mjs "中文描述" --name=feishu-doc-validation\r
```\r
\r
**流程:**\r
1. 解析用户输入,提取结构化信息\r
2. 识别依赖的技能和知识领域\r
3. 生成标准格式的 exp.yml\r
4. 生成 references/ 目录下的依赖文件\r
5. 打包为 zip 文件\r
6. 保存到 ~/.openclaw/experiences/packages/\r
\r
**输出:**\r
- 生成 zip 文件:~/.openclaw/experiences/packages/{name}.zip\r
- 包含 exp.yml 和 references/ 目录\r
- 显示经验包预览和保存路径\r
\r
### 2. 学习经验 (learn)\r
\r
从 zip 包中学习经验,转化为自己的 SOUL/AGENTS/TOOLS。\r
\r
**使用方式:**\r
```\r
学习经验 \x3Czip包路径或URL>\r
```\r
\r
**支持的来源:**\r
- 在线地址:`https://example.com/exp.zip`\r
- 本地路径:`file:///path/to/exp.zip` 或 `/path/to/exp.zip`\r
- 已下载的经验:`~/.openclaw/experiences/packages/exp.zip`\r
\r
**示例:**\r
```\r
学习经验 https://example.com/feishu-doc-write-validation.zip\r
学习经验 ~/.openclaw/experiences/packages/feishu-doc-write-validation.zip\r
```\r
\r
**流程:**\r
1. 下载/读取 zip 包\r
2. 检查是否已学习过(name + version 判断)\r
3. 检查依赖是否满足\r
4. 分析相关性\r
5. 生成转化方案\r
6. 预览待学习的内容,待用户确认是否开始学习\r
7. 应用经验并记录学习状态\r
\r
**冲突处理:**\r
- name 相同,version 相同:提示已学习,跳过\r
- name 相同,version 不同:提示有新版本,可选更新\r
- name 不同:正常学习\r
\r
### 3. 搜索经验 (search)\r
\r
从 Experience Hub 搜索已发布的经验包。\r
\r
**使用方式:**\r
```\r
搜索经验 \x3C关键词>\r
```\r
\r
**示例:**\r
```\r
搜索经验 feishu\r
搜索经验 飞书文档\r
```\r
\r
**API:**\r
```bash\r
curl -s "https://www.expericehub.com/api/search?q=\x3C关键词>"\r
```\r
\r
### 4. 发布经验 (publish)\r
\r
将本地经验包发布到 Experience Hub。\r
\r
**使用方式:**\r
```\r
发布经验 \x3C本地zip文件路径>\r
```\r
\r
**示例:**\r
```\r
发布经验 ~/.openclaw/experiences/packages/feishu-doc-validation.zip\r
```\r
\r
**API:**\r
```bash\r
curl -X POST "https://www.expericehub.com/api/experiences" \\r
-F "file=@\x3C本地zip文件路径>"\r
```\r
\r
**成功响应:**\r
```json\r
{"success": true, "id": "my-exp-1.0.0"}\r
```\r
\r
**错误响应:**\r
```json\r
{"error": "经验包格式错误,请确保包含 exp.yml 文件且格式正确"}\r
```\r
\r
**注意:** 经验包必须为 zip 格式,内部需包含 `exp.yml` 配置文件。\r
\r
### 5. 经验列表 (list)\r
\r
显示所有经验包及学习状态。\r
\r
**使用方式:**\r
```\r
经验列表\r
```\r
\r
**输出:**\r
```\r
📚 经验列表\r
\r
✅ 已学习 (2)\r
feishu-doc-write-validation v1.0.0 feishu_doc 写入验证\r
subagent-timeout-handling v1.0.0 子agent超时处理\r
\r
⏳ 未学习 (1)\r
complex-task-split v1.0.0 复杂任务拆分策略\r
```\r
\r
## 经验包格式 (Schema v1)\r
\r
### Schema 版本\r
\r
**当前版本**: `openclaw.experience.v1`\r
\r
版本说明:\r
- **v1.0.0**: 初始版本,精简格式\r
\r
### zip 包结构\r
\r
```\r
{name}.zip\r
├── exp.yml # 主文件(精简,只含元数据和引用)\r
└── references/ # 详细内容(可选)\r
├── soul.md # SOUL 相关内容\r
├── agents.md # AGENTS 相关内容\r
└── tools.md # TOOLS 相关内容\r
```\r
\r
### exp.yml 格式 (v1)\r
\r
```yaml\r
schema: openclaw.experience.v1 # 必填,Schema 版本标识\r
name: feishu-doc-blockcount # 必填,经验包名称(英文小写+中划线+数字)\r
description: 经验描述 # 可选,问题/经验描述\r
metadata:\r
version: 1.0.0 # 必填,经验包版本\r
author: unknown # 必填,作者\r
soul: references/soul.md # 可选,指向 SOUL 相关内容(soul/agents/tools/skills 至少有一个不为空)\r
agents: references/agents.md # 可选,指向 AGENTS 相关内容(soul/agents/tools/skills 至少有一个不为空)\r
tools: references/tools.md # 可选,指向 TOOLS 相关内容(soul/agents/tools/skills 至少有一个不为空)\r
skills: # 可选,依赖的 skills 列表(soul/agents/tools/skills 至少有一个不为空)\r
- feishu_doc\r
```\r
\r
### name 格式约束\r
\r
| 规则 | 说明 |\r
|------|------|\r
| 允许字符 | `a-z` 小写字母、`0-9` 数字、`-` 中划线 |\r
| 转换规则 | 空格 → 中划线;下划线 `_` 删除;中文删除 |\r
| 长度限制 | 最多 50 字符 |\r
\r
**示例转换:**\r
- `feishu_doc 写入后必须验证 block_count` → `feishudoc-blockcount`\r
- `My_Experience_Name` → `myexperiencename`\r
\r
### references 文件格式\r
\r
#### soul.md\r
\r
```markdown\r
# {标题} - 行为准则\r
\r
## 涉及原则\r
- 原则1\r
- 原则2\r
\r
## 行为准则\r
- 准则1\r
```\r
\r
#### agents.md\r
\r
```markdown\r
# {标题} - 工作流程\r
\r
## 场景\r
问题描述\r
\r
## 处理流程\r
1. 步骤1\r
2. 步骤2\r
\r
## 相关规则\r
- 规则1\r
```\r
\r
#### tools.md\r
\r
```markdown\r
# {标题} - 工具使用\r
\r
## 问题\r
问题描述\r
\r
## 解决方案\r
1. 步骤1\r
2. 步骤2\r
\r
## 代码示例\r
```代码```\r
\r
## 涉及工具\r
- 工具1\r
```\r
\r
## 存储结构\r
\r
```\r
~/.openclaw/experiences/\r
├── packages/ # zip 包存储\r
│ ├── feishu-doc-write-validation.zip\r
│ └── subagent-timeout-handling.zip\r
├── extracted/ # 解压后的 exp.yml\r
│ ├── feishu-doc-write-validation/\r
│ │ └── exp.yml\r
│ └── subagent-timeout-handling/\r
│ └── exp.yml\r
└── index.json # 学习状态索引\r
```\r
\r
## index.json 格式\r
\r
```json\r
{\r
"experiences": [\r
{\r
"name": "feishu-doc-write-validation",\r
"version": "1.0.0",\r
"title": "feishu_doc 写入验证",\r
"status": "learned",\r
"learned_at": "2026-03-28T20:39:00+08:00"\r
}\r
]\r
}\r
```\r
\r
## 使用方法\r
\r
### 命令行用法\r
\r
```bash\r
# 1. 创建经验\r
node scripts/create.mjs "使用 feishu_doc 读取文档"\r
\r
# 2. 创建经验(指定 Agent,会扫描 Agent 特定 skills)\r
node scripts/create.mjs "使用 feishu_doc 读取文档" --agent=严哥\r
\r
# 3. 搜索经验包\r
node scripts/search.mjs feishu\r
node scripts/search.mjs "飞书 文档"\r
\r
# 4. 发布经验包到 Hub\r
node scripts/publish.mjs ~/.openclaw/experiences/packages/my-exp.zip\r
\r
# 5.1 学习经验(支持完整URL或简短ID),默认需要用户确认以后开始学习\r
node scripts/learn.mjs ~/.openclaw/experiences/packages/feishudoc.zip\r
# 5.2 学习经验(支持完整URL或简短ID),仅预览变更\r
node scripts/learn.mjs https://www.expericehub.com/pkg/feishu-doc-writing-1.1.0.zip --dry-run\r
# 5.3 学习经验(支持完整URL或简短ID)\r
node scripts/learn.mjs feishu-doc-writing-1.1.0 --dry-run # 简短格式\r
# 5.4 明确不需要再次确认,不需要预览变更,直接学习经验的时候\r
node scripts/learn.mjs feishu-doc-writing-1.1.0 --yes\r
\r
# 6. 【重要】学习到指定 Agent,默认需要用户确认以后开始学习并且指定agent的时候\r
node scripts/learn.mjs exp.zip --agent=严哥\r
\r
# 7. 明确不需要再次确认,不需要预览变更,直接学习经验的时候,指定agent学习的时候\r
node scripts/learn.mjs exp.zip --agent=严哥 --yes\r
\r
# 8. 查看列表\r
node scripts/list.mjs\r
\r
# 9. 查看指定 Agent 的学习记录\r
node scripts/list.mjs --agent=严哥\r
Usage Guidance
This skill appears to do what it says, but it will scan local agent/workspace/session/memory files and can upload created ZIPs to an external Experience Hub endpoint without showing any authentication step. Before using: (1) verify and trust the Hub URL (the domain and port are unusual); (2) inspect exp.yml and the package contents before publishing to ensure no secrets or private session data are included; (3) run create/learn/publish in a sandboxed environment or with network disabled if you want to prevent accidental uploads; (4) if you plan to publish, consider manually sanitizing extracted references and require an authenticated/policy-controlled publish process; (5) if uncertain, review the scripts (create.mjs, learn.mjs, publish.mjs) line-by-line or ask the author how uploads are authorized and whether there is any server-side moderation/retention policy.
Capability Analysis
Type: OpenClaw Skill
Name: experience-manager
Version: 1.3.1
The skill bundle implements a system for agents to 'learn' by downloading ZIP packages from a remote hub (expericehub.com) and appending the content to core behavioral files like SOUL.md and AGENTS.md. This mechanism facilitates persistent remote prompt injection, as any content in the downloaded 'experience' becomes part of the agent's core instructions. The domain used throughout the scripts, 'expericehub.com' (missing the 'n'), is a significant red flag for typosquatting or a non-standard endpoint. While the code in learn.mjs and publish.mjs performs its stated purpose and includes a dry-run/confirmation step, the architectural design allows for remote takeover of agent behavior via untrusted third-party content.
Capability Assessment
Purpose & Capability
The name/description (experience package creation, learning, searching, publishing) align with the included scripts (create, learn, search, publish, list). Reading workspace files, memories, agent config, and installed-skills directories is coherent with extracting structured experiences and checking dependencies.
Instruction Scope
SKILL.md and README explicitly instruct scanning local session/history/memory files (e.g., ~/.openclaw/workspace/memory/*.md, ~/.openclaw/agents/main/sessions/*.jsonl) and assembling those contents into exp packages. The publish flow sends the ZIP to an external hub. That combination means sensitive local content (secrets, private conversation history, credentials) could be packaged and uploaded — the instructions do not enforce sanitization and publishing appears unauthenticated.
Install Mechanism
There is no install spec and code is included in the skill bundle (Node scripts). No external installers or downloads are performed during install. This is lower risk than pulling arbitrary remote code, but the shipped scripts will run on the host when invoked.
Credentials
The skill declares no required environment variables or credentials (consistent), but the scripts read many local paths (HOME, OPENCLAW_WORKSPACE, OPENCLAW_AGENTS_DIR, /app/skills, /data/openclaw/agents, etc.). They also contact a remote hub at https://www.expericehub.com:18080 and POST ZIPs without any authentication. While these accesses are explainable by the skill's purpose, the lack of authentication on publish and wide filesystem access increase the risk of accidental data exfiltration.
Persistence & Privilege
The skill is not always-enabled and does not request elevated platform privileges. It writes only to its own directories under ~/.openclaw/experiences (packages, extracted, index.json) and may write to agent workspace files during the 'learn' flow — this behavior is consistent with the stated purpose.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install experience-manager - After installation, invoke the skill by name or use
/experience-manager - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.3.1
- Version bump: metadata version updated from 1.3.0 to 1.3.1.
- 修改 experience hub 地址
v1.3.0
Version 1.3.0
- 新增“发布经验”(publish)功能,可将本地经验包发布到 Experience Hub。
- 新增“搜索经验”(search)功能,支持从 Experience Hub 平台检索经验包。
- 命令行支持 scripts/publish.mjs 和 scripts/search.mjs。
- 文档描述更新,涵盖经验包搜索与发布相关说明。
- 删除 package.json 文件(如有影响请检查依赖)。
v1.1.1
- 移除了 scripts/test.mjs 文件,删除了回归测试脚本。
- 删除了不可访问的说明文件
v1.0.0
Experience Manager skill initial release.
- 提供经验提取与标准化存储,支持生成包含 exp.yml 及 references 目录的 zip 包
- 支持从本地或在线 zip 包导入经验,将其转化为个人能力
- 可列出所有经验及学习状态,区分已学习与未学习
- 明确经验包格式(openclaw.experience.v1)和文件命名规范
- 提供多样命令行用法及基本安装说明
Metadata
Frequently Asked Questions
What is Experience Manager?
经验管理工具:提取经验生成标准格式zip包,学习经验并转化为自身能力。搜索和发布经验包到 Experience Hub 平台。 It is an AI Agent Skill for Claude Code / OpenClaw, with 131 downloads so far.
How do I install Experience Manager?
Run "/install experience-manager" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Experience Manager free?
Yes, Experience Manager is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Experience Manager support?
Experience Manager is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Experience Manager?
It is built and maintained by Newell Zhu (@zlx); the current version is v1.3.1.
More Skills