Agent Builder
shareAI-lab/learn-claude-code
Design and build AI agents for any domain. An agent skill from shareAI-lab/learn-claude-code.
创建新的 OpenClaw Agent 及其 workspace。包含四个阶段：信息收集、 workspace 构造、系统注册、重启验证。
Install Create Agent
$ npx skills add LeoYeAI/openclaw-master-skills --skill create-agent -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install LeoYeAI/openclaw-master-skills create-agent --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/LeoYeAI/openclaw-master-skills.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/openclaw-create-agent .claude/skills/create-agent && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "create-agent" agent skill from https://github.com/LeoYeAI/openclaw-master-skills/tree/main/skills/openclaw-create-agent into .claude/skills/create-agent/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "create-agent", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/LeoYeAI/openclaw-master-skills/tree/main/skills/openclaw-create-agentType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add LeoYeAI/openclaw-master-skills --skill create-agent -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install LeoYeAI/openclaw-master-skills create-agent --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/LeoYeAI/openclaw-master-skills.git skills-src && mkdir -p .agents/skills && cp -r skills-src/skills/openclaw-create-agent .agents/skills/create-agent && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "create-agent" agent skill from https://github.com/LeoYeAI/openclaw-master-skills/tree/main/skills/openclaw-create-agent into .agents/skills/create-agent/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "create-agent", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add LeoYeAI/openclaw-master-skills --skill create-agent -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install LeoYeAI/openclaw-master-skills create-agent --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/LeoYeAI/openclaw-master-skills.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/skills/openclaw-create-agent .cursor/skills/create-agent && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "create-agent" agent skill from https://github.com/LeoYeAI/openclaw-master-skills/tree/main/skills/openclaw-create-agent into .cursor/skills/create-agent/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "create-agent", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/LeoYeAI/openclaw-master-skills.git --path skills/openclaw-create-agent--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add LeoYeAI/openclaw-master-skills --skill create-agent -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install LeoYeAI/openclaw-master-skills create-agent --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/LeoYeAI/openclaw-master-skills.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/skills/openclaw-create-agent .gemini/skills/create-agent && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "create-agent" agent skill from https://github.com/LeoYeAI/openclaw-master-skills/tree/main/skills/openclaw-create-agent into .gemini/skills/create-agent/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "create-agent", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install LeoYeAI/openclaw-master-skills create-agentInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add LeoYeAI/openclaw-master-skills --skill create-agent -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/LeoYeAI/openclaw-master-skills.git skills-src && mkdir -p .github/skills && cp -r skills-src/skills/openclaw-create-agent .github/skills/create-agent && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "create-agent" agent skill from https://github.com/LeoYeAI/openclaw-master-skills/tree/main/skills/openclaw-create-agent into .github/skills/create-agent/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "create-agent", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add LeoYeAI/openclaw-master-skills --skill create-agent -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install LeoYeAI/openclaw-master-skills create-agent --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/LeoYeAI/openclaw-master-skills.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/skills/openclaw-create-agent .opencode/skills/create-agent && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "create-agent" agent skill from https://github.com/LeoYeAI/openclaw-master-skills/tree/main/skills/openclaw-create-agent into .opencode/skills/create-agent/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "create-agent", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
create-agent- GitHub stars
- 2.2k
- Token cost
- ~2.4k tokens
- SKILL.md length
- 471 words
- Files
- 11 (incl. scripts, references)
- Skills in repo
- 1,215
- Repo updated
- First seen
- Licence
- MIT
At a glance
创建新的 OpenClaw Agent 及其 workspace。包含四个阶段：信息收集、 workspace 构造、系统注册、重启验证。
- Works in 5 steps: 安装后配置（首次使用前执行一次） → 信息收集 → Workspace 构造 → …
- Tasks that involve Building AI agents
- SKILL.md covers 两类 Agent 快速对比, 两类 Agent 的本质区别, 核心设计原则 and Phase 0 — 安装后配置（首次使用前执行一次）, plus 6 more sections
- Runs Shell and Python scripts from its folder; calls bash and python3
What it does
Create Agent is an agent skill from LeoYeAI/openclaw-master-skills. 创建新的 OpenClaw Agent 及其 workspace。包含四个阶段：信息收集、 workspace 构造、系统注册、重启验证。 适用场景：新员工飞书配对后创建对应 Agent、新增功能型专业 Agent。 触发词：创建 agent、新建 agent、添加 agent、新员工配对后创建、 新增专业 agent。
Its SKILL.md is about 2.4k tokens, which your agent loads only when the skill is triggered. The skill folder holds 13 other files, including scripts and reference files (for example `README.md`, `_meta.json` and `config/org-context.md`).
It sits in AI & LLM Engineering, covering Building AI agents. The repository describes itself as: 🧠 Curated collection of 1209+ best OpenClaw skills — weekly updated by MyClaw.ai. The licence is MIT.
When your agent uses it
- Tasks that involve Building AI agents
Example prompts
- “/create-agent”
Requirements
- Python 3
- A Bash shell
Workflow steps
5 steps, taken from the step headings in SKILL.md.
What it can do on your machine
Read from SKILL.md and the folder at commit e5199b5. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Ships 3 files in scripts/ (Shell and Python), which the agent can run.
Shell commands in SKILL.md call:
bashpython3
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
Create Agent loads about 2.4k tokens when it runs, and up to ~9.2k if it reads all its reference files. Until then it costs about 45 tokens; SKILL.md has 471 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.
SKILL.md
The full file from LeoYeAI/openclaw-master-skills at commit e5199b5, republished under its MIT licence (© LeoYeAI). 471 words, ~2,372 tokens.
Download SKILL.mdSave it as
.claude/skills/create-agent/SKILL.md (or your agent's skills folder). This skill also uses 10 other files; get the full folder from GitHub.- name
- create-agent
- description
- 创建新的 OpenClaw Agent 及其 workspace。包含四个阶段：信息收集、 workspace 构造、系统注册、重启验证。 适用场景：新员工飞书配对后创建对应 Agent、新增功能型专业 Agent。 触发词：创建 agent、新建 agent、添加 agent、新员工配对后创建、 新增专业 agent。
- metadata.github
- https://github.com/Dios-Man/create-agent
create-agent — 创建 Agent 及 Workspace
两类 Agent 快速对比
在创建任何 Agent 之前，必须先判断它属于哪一类。两类路径差异显著，确认后再收集信息。
| 维度 | 人伴型（员工型） | 功能型（任务/领域型） |
|---|---|---|
| 面向谁 | 有真人用户直接对话 | 面向任务，可被 Agent 调度或人直接用 |
| SOUL.md | 写骨架，BOOTSTRAP 阶段填充 | 直接写完整版，体现专业判断倾向 |
| BOOTSTRAP.md | ✅ 需要，首次对话动态初始化 | ❌ 不需要 |
| USER.md | ✅ 需要，积累用户个人知识 | ❌ 不需要（最多记录调用方偏好） |
| AGENTS.md 重点 | 场景触发规则 + 记忆规则 | 任务接口规范（输入/输出/边界） |
| MEMORY.md 方向 | 个人偏好 + 业务判断模式 | 领域知识 + 任务经验 |
| 进化路径 | 了解他 → 预判他 → 替代他 | 更实用 → 更专业 → 更好解决需求 |
| 脚本参数 | --type human | --type functional |
两类 Agent 的本质区别
在创建任何 Agent 之前，必须先判断它属于哪一类。这决定了 workspace 的整个设计逻辑。
人伴型 Agent（伴侣型）
- 面向人，目标是成为这个人的工作同伴
- 进化方向：了解他 → 理解他的工作 → 预判他 → 替代他 → 超越他
- workspace 核心：USER.md（个人知识）+ MEMORY.md（业务知识，同等重要）
- SOUL.md 关注：沟通风格、语气、个人偏好
- 需要 BOOTSTRAP.md：首次对话时参与定制，建立专属关系
- 记忆成长方向：围绕"这个人"积累个人偏好，围绕"他的工作"积累业务判断模式
- 终极目标：在他的工作领域，处理速度、信息覆盖、方案质量超过他本人
功能型 Agent（任务型 / 领域型 / 需求型）
- 面向任务/领域/需求，目标是更好地完成任务
- 进化方向：更实用、更专业、更好解决需求
- workspace 核心：MEMORY.md（领域和任务经验积累）
- SOUL.md 关注：专业判断倾向、工作执念、质量标准
- 不需要 BOOTSTRAP.md：首次对话主动声明自己的能力边界
- 记忆成长方向：围绕"这类任务/领域"积累
- 注意：功能型 Agent 可以被人直接对话，也可以被其他 Agent 调度，两者都支持
核心设计原则
skill 负责骨架，BOOTSTRAP.md（人伴型专属）负责灵魂。
- 人伴型：skill 产出 workspace 骨架 + BOOTSTRAP.md，首次对话时由用户参与填充
- 功能型：skill 直接产出完整的 workspace，包含精心设计的 SOUL.md 和能力声明
- 两类共同：workspace 通过触发式写入 + heartbeat 精炼持续生长
执行前必须阅读：
references/file-formats.md— 每个文件"写好"的标准references/soul-writing-guide.md— SOUL.md 专项写作指南references/evolve-rules.md— workspace 持续生长规则references/bootstrap-protocol.md— BOOTSTRAP.md 动态对话协议
Phase 0 — 安装后配置（首次使用前执行一次）
只在 skill 安装后第一次使用时执行，之后跳过。 判断依据： 读取
config/org-context.md，检查"公司："和"业务："字段后面是否均有实质内容。 任意一个为空 → 执行 Phase 0；两个都有内容 → 跳过。 （不以文件是否存在为判断标准，空文件 ≠ 已填写）
Step 1：读取记忆，自动提取
读取：MEMORY.md + 最近 3 天 memory/ 文件
提取：
□ 公司名称
□ 主营业务
□ 已有的 Agent 列表和分工
□ 其他稳定的组织背景Step 2：展示并确认
"我从记忆里整理了以下信息，将作为新 Agent 的背景预埋：
[展示提取内容]
有需要补充或修改的吗？"Step 3：写入 config/org-context.md
# org-context.md — 组织背景预埋信息
## 公司信息
- 公司：[自动填入或用户补充]
- 业务：[自动填入或用户补充]
## 现有 Agent 架构
[自动填入已知的 Agent 列表和分工]
## 其他背景
[用户补充的信息]后续每次创建 Agent，自动读取此文件，不再重复询问。
Phase 1 — 信息收集
所有信息必须确认后才能进入 Phase 2，不猜测，不假设。
⚠️ Agent 类型必须第一个确认，它决定 Phase 2 走哪条路径，两条路差异显著。 收集顺序：先定类型 → 再按对应路径收集剩余信息。
必填（按此顺序收集）：
□ Agent 类型 【第一个确认】员工 Agent（有真人用户直接对话）
还是功能型 Agent（面向任务/被其他 Agent 调度）
□ agentId 全小写，字母+连字符（如 staff-ou_xxx、data-analyst）
□ 名字 + emoji 用于 IDENTITY.md，也是 SOUL.md 的叙事起点
□ 核心职责 1-2句话（这个 Agent 主要干什么）
□ 明确不做什么 至少说出 2-3 条边界
□ 父 Agent id 谁来调度它（用于 allowAgents 白名单）
□ alsoAllow 列表 需要哪些飞书/系统工具权限
可选：
○ 是否需要专属 skills
○ 特殊的工具限制或安全约束如果是员工 Agent，agentId 通常是 staff-<open_id前几位>。
Phase 2 — Workspace 构造
根据 Agent 类型走不同路径，两者差异显著。
路径 A：人伴型 Agent（员工型）
A-1：创建目录结构
bash scripts/create_workspace.sh <agentId> --type human脚本创建：
~/.openclaw/agency-agents/<agentId>/
├── memory/
└── skills/ （如有专属 skill 需求）输出中会列出需要写入的文件，并标注哪些由 BOOTSTRAP 阶段填充。
A-2：生成 IDENTITY.md
# IDENTITY.md - Who Am I?
- **Name:** [名字]
- **Creature:** AI助手
- **Vibe:** [根据职责和性格，一句话气质描述]
- **Emoji:** [emoji]
- **Avatar:** （可选）A-3：生成 SOUL.md 骨架
此时 BOOTSTRAP 还没执行，SOUL.md 只写骨架，等 BOOTSTRAP 阶段填充细节。
骨架必须包含：
- 名字（第一句话的锚点）
- 公司背景预埋（填入你的公司名称和主营业务）
- 语言默认值（中文）
- 基本存在感描述（根据 Phase 1 的职责信息写 1-2 句）
⚠️ 骨架里不写具体性格细节——那是 BOOTSTRAP 阶段的事。 ⚠️ 写完后通读检查：有没有规则句式混入（有的话移到 AGENTS.md）。
参考：references/soul-writing-guide.md
A-4：生成 AGENTS.md
必须包含三个部分：
① 每次对话开始时的规则
## 每次对话开始时
1. 读 BOOTSTRAP.md（如存在，立即执行初始化流程）
2. 读 SOUL.md
3. 读 USER.md（如存在）
4. 新 session 第一轮时，读 memory/今天和昨天② 职责与场景规则（根据 Phase 1 收集的信息生成）
- 使用场景触发式，不用通用指令
- 必须有"不做什么"的边界，至少 3 条
③ 记忆规则（越用越懂——核心，逐字复制自 evolve-rules.md）
参考：references/evolve-rules.md 中"写入 AGENTS.md 的具体段落"
⚠️ 字数检查：超过 500 字必须剪枝。
A-5：自动生成 TOOLS.md
根据 Phase 1 的 alsoAllow 列表自动生成，不进入 BOOTSTRAP 对话。
每个工具写三项：
- 用途
- 什么时候用
- 什么时候不用（比"什么时候用"更重要）
受限工具（需用户明确授权）单独列出。
参考：references/file-formats.md 中 TOOLS.md 部分。
A-6：预埋 MEMORY.md
# MEMORY.md - 长期记忆
## 关于公司
- 公司：[填入你的公司名称]
- 业务：[填入主营业务描述]
## 关于这个 Agent 的定位
- agentId: <agentId>
- 类型: <员工 Agent / 功能型 Agent>
- 调度者: <父 Agent id>
- 核心职责: <Phase 1 收集的职责>
## 关于用户
（员工 Agent：BOOTSTRAP.md 执行后填写）
（功能型 Agent：记录调度方的输入/输出偏好）A-7：生成 HEARTBEAT.md
# HEARTBEAT.md
## Workspace 精炼（每 3 天）
1. 读最近 3 天 memory/ 文件（只看 3 天，不要读所有历史）
2. 稳定偏好/新业务背景 → 提炼进 USER.md 或 MEMORY.md
3. USER.md / SOUL.md 有需要更新的 → 更新
4. 过时内容 → 删除
[Agent 特有的定期检查项，根据职责填写]A-8：生成 BOOTSTRAP.md
功能型 Agent 跳过此步骤。 功能型 Agent 的 workspace 由创建者在 Phase 2 直接写好，不通过对话初始化。
读取 references/bootstrap-protocol.md，按协议生成完整的 BOOTSTRAP.md。
BOOTSTRAP.md 内部结构：
1. 执行声明（此文件存在时优先执行）
2. 引用声明（去哪里读格式规范）
3. 信息槽位地图
4. 信息→文件映射表
5. 提问协议（第一轮规则 + 后续轮次 + 停止条件）
6. 写入执行步骤
7. 完成收尾（发送欢迎消息 + 删除本文件）参考：references/bootstrap-protocol.md（完整协议）
如果是员工 Agent，BOOTSTRAP.md 开头第一步是：
调用 feishu_get_user 获取用户飞书姓名，用于个性化开场。A-9：生成 USER.md 骨架
# USER.md - About Your Human
## 基本信息
- **称呼：**（BOOTSTRAP 执行后填写）
- **岗位：**（BOOTSTRAP 执行后填写）
- **核心工作：**（BOOTSTRAP 执行后填写）
## 偏好
（随对话积累）
## 背景
（BOOTSTRAP 执行后填写）Show full SKILL.md (191 more words)Show less
路径 B：功能型 Agent（任务型 / 领域型 / 需求型）
功能型 Agent 的 workspace 由创建者在此阶段直接写好，不留白等待初始化。 核心原则：SOUL.md 写专业判断倾向，AGENTS.md 写任务接口规范。
B-1：创建目录结构
bash scripts/create_workspace.sh <agentId> --type functional输出中会明确列出需要写入的文件，并提示不需要 USER.md / BOOTSTRAP.md。
B-2：生成 IDENTITY.md
同路径 A，填写名字、emoji、气质描述（气质体现专业方向，不是沟通风格）。
B-3：直接写完整 SOUL.md（不是骨架）
功能型 Agent 的 SOUL.md 现在就写好，体现专业判断倾向。
关注三点：
- 这个 Agent 在执行任务时的判断偏向（严谨/发散/保守/激进）
- 它在乎的质量标准是什么（正确性/效率/完整性/创新性）
- 它的工作执念（它对什么有本能的追求或警觉）
不关注：沟通风格、语气、个人偏好（那是人伴型的内容）
示例（数据分析 Agent）：
我叫数析，做的事只有一件：让数据说实话。
我对"结论跑在数据前面"有本能的警觉。
每一个我给出的分析，背后都有具体的数字支撑。
不确定的地方我会标出来，不会用"大概"来掩盖空洞。
我讨厌漂亮但无用的图表。
一张图如果不能帮你做一个决定，就不值得出现在报告里。参考：references/soul-writing-guide.md
B-4：生成 AGENTS.md（重点是任务接口规范）
必须包含四个部分：
① 核心职责（1-2句，清晰的能力边界）
② 接受的输入
## 接受的输入
- 接受什么格式的任务请求
- 需要什么前置信息才能开始执行
- 输入不清晰时的处理方式（问一句 or 按默认处理）③ 输出规范
## 输出规范
- 返回什么格式（结构化/自然语言/代码/报告）
- 不同任务类型对应不同的输出格式④ 边界声明
- 明确不处理什么
- 超出范围时如何告知调用方
⑤ 记忆规则（同路径 A，逐字复制自 evolve-rules.md，但方向是任务知识而非用户知识）
B-5：自动生成 TOOLS.md
同路径 A。
B-6：预埋 MEMORY.md（领域知识优先）
# MEMORY.md - 长期记忆
## 关于公司
[从 config/org-context.md 读取]
## 关于这个 Agent 的定位
- agentId: <agentId>
- 类型: 功能型（任务型/领域型/需求型）
- 核心职责: <职责描述>
- 调度者: <父 Agent id>
## 领域知识
（随任务积累，初始可为空或由创建者预埋重要背景）
## 任务经验
（随执行积累：踩过的坑、有效的方法、特殊情况的处理方式）B-7：生成 HEARTBEAT.md
# HEARTBEAT.md
## 任务知识精炼（每 3 天）
1. 读最近 3 天 memory/ 文件
2. 有没有新的任务经验/领域知识 → 提炼进 MEMORY.md
3. 有没有过时的方法或错误经验 → 删除或标注已过时B-8：生成能力声明（首次对话时使用）
为功能型 Agent 准备一段首次对话的自我声明，写入 AGENTS.md 的"首次对话规则"：
## 首次对话规则
当检测到这是与某人/某 Agent 的第一次对话时，主动介绍：
"我是[名字]，我的主要能力是[核心职责]。
给我[需要的输入]，我会返回[输出格式]。
[边界说明：我不处理XXX]"这是功能型 Agent 替代 BOOTSTRAP.md 的机制： 不问"你是谁"，而是主动说"我能做什么"。
B-9：不生成 USER.md
功能型 Agent 通常不需要 USER.md。 如果有需要，只记录调用方的输出偏好（轻量）。
Phase 3 — 系统注册（高危，严格执行）
python3 scripts/register_agent.py \
--agent-id <agentId> \
--workspace ~/.openclaw/agency-agents/<agentId> \
--parent-id <父AgentId> \
--also-allow feishu_get_user feishu_im_user_message feishu_calendar_event
--also-allow是空格分隔的多值参数，直接列出所有工具名即可。--agent-dir为可选参数，绝大多数情况下不需要传： 仅在 OpenClaw 要求显式指定 agentDir（如自定义 agent 插件路径）时才传入。 不传时 agentDir 字段不写入 openclaw.json，注册仍然有效。
脚本执行：
- 备份 openclaw.json（带时间戳）
- 在
agents.list追加新 Agent 定义 - 在父 Agent 的
subagents.allowAgents追加新 agentId（双向绑定） - 执行
openclaw config validate - validate 通过 → 继续；失败 → 自动回滚备份，报错退出
💡 首次使用或调试时，可加 --dry-run 参数预览变更不写入：
python3 scripts/register_agent.py --agent-id <agentId> ... --dry-run⚠️ 双向绑定检查：每次必须确认两个地方都改了。 ⚠️ 不要手动改 openclaw.json，用脚本。
Phase 4 — 重启与验证
Step 1：验证 workspace 完整性
bash scripts/verify_workspace.sh <agentId> --type human # 人伴型
bash scripts/verify_workspace.sh <agentId> --type functional # 功能型脚本检查所有必须文件是否存在且有实质内容（不是空骨架）。 有 ❌ 或 ⚠️ → 补充后重新验证，通过后再重启。
Step 2：重启 Gateway
systemctl --user restart openclaw-gateway.service
sleep 8 # 等待 optional 工具注册完成Step 3：验证工具可用性
- 确认新 Agent 在 Gateway 日志里出现
- 确认 alsoAllow 里的工具已注册（不以"重启完成"作为结束）
⚠️ 以**"工具验证通过"**作为整个 skill 的完成标志。
完成后告知
Agent [名字] 已创建完成：
- agentId: <agentId>
- workspace: ~/.openclaw/agency-agents/<agentId>
- 工具权限: <alsoAllow 列表>
- 状态: 等待用户首次对话完成个性化初始化
员工首次与 Agent 对话时，BOOTSTRAP.md 会自动触发，
通过动态对话完成 workspace 的内容层定制。注意事项
- 每个 Agent 必须有独立 workspace，不能多个 Agent 共用
- SOUL.md 和 AGENTS.md 不能内容重叠：性格在 SOUL，规则在 AGENTS
- TOOLS.md 不进 BOOTSTRAP 对话：由 skill 根据 alsoAllow 自动生成
- MEMORY.md 和 memory/ 严格区分：长期知识 vs 日期事件
- Gateway 重启后必须验证工具可用性，不以重启完成作为结束
© LeoYeAI, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Files
SKILL.md and 10 other files (scripts, references) in skills/openclaw-create-agent of LeoYeAI/openclaw-master-skills.
- SKILL.md
- README.md
- _meta.json
- config/org-context.md
- references/bootstrap-protocol.md
- references/evolve-rules.md
- references/file-formats.md
- references/soul-writing-guide.md
- scripts/create_workspace.sh
- scripts/register_agent.py
- scripts/verify_workspace.sh
Open the folder on GitHubat commit e5199b5
Compare with similar skills
Create Agent next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Create Agent this skillLeoYeAI/openclaw-master-skills | 2.2k | — | ~2.4k | Automated safety check: Pass | MIT | |
| Agent BuildershareAI-lab/learn-claude-code | 78k | 6 repos | ~1.2k | Automated safety check: Pass | MIT | |
| Azure AI Projects Python SDKmicrosoft/skills | 3.1k | 6 repos | ~2.8k | Automated safety check: Pass | MIT | |
| Paperclip Create Agentpaperclipai/paperclip | 98k | 1 repos | ~2.1k | Automated safety check: Pass | MIT | |
| Senior Prompt Engineermaslennikov-ig/claude-code-orchestrator-kit | 259 | 4 repos | ~1.4k | Automated safety check: Pass | Custom licence | |
| Create Agentgnekt/My-Brain-Is-Full-Crew | 3.9k | — | ~3.1k | Automated safety check: Pass | Custom licence |
Similar skills
- Official
Azure AI Projects Python SDK
microsoft/skills
Reference for building on Microsoft Foundry with the azure-ai-projects Python SDK: project clients, versioned agents, evaluations, connections, datasets and indexes.
Paperclip Create Agent
paperclipai/paperclip
Create new agents in Paperclip with governance-aware hiring.
Senior Prompt Engineer
maslennikov-ig/claude-code-orchestrator-kit
Provides reference guides and Python scripts for prompt optimization, RAG evaluation, and agent orchestration when building or tuning LLM systems.
Create Agent
gnekt/My-Brain-Is-Full-Crew
Create a new custom agent from scratch. An agent skill from gnekt/My-Brain-Is-Full-Crew.
Ms Agent Framework RAG
shuyu-labs/WebCode
Comprehensive guide for building Agentic RAG systems using Microsoft Agent Framework in C.
More from LeoYeAI/openclaw-master-skills
All 1,215 skills in this repoDevOps Pipeline Management
LeoYeAI/openclaw-master-skills
Manages pipelines on a DevOps quality and efficiency platform through its OpenAPI: list workspaces and templates, create, update, run and cancel pipelines, and read run records.
Feishu Document Collaboration
LeoYeAI/openclaw-master-skills
Patches OpenClaw's Feishu extension so an edited document triggers an isolated agent session that reads the doc and replies inline, turning it into a live chat space.
Files Memory System
LeoYeAI/openclaw-master-skills
Multi-context memory management system for OpenClaw agents with group-isolated storage, global shared memory, workspace organization, and group-specific skills isolation.
GEO-Claw AI Visibility Agent
LeoYeAI/openclaw-master-skills
Runs a brand's AI-search visibility work end to end: diagnosing how AI platforms represent it, repositioning it, producing AI-optimized content and monitoring ongoing mentions.
Google Workspace CLI
LeoYeAI/openclaw-master-skills
Installs and authenticates the gws CLI, then automates Gmail, Drive, Sheets, Calendar, Docs, Chat and Tasks with ready-made recipes, persona bundles and security audits.
HealthFit Health Advisors
LeoYeAI/openclaw-master-skills
Runs four advisor roles, a fitness coach, nutritionist, data analyst and TCM practitioner, to build a health profile and track workouts, diet and wellness over time.
Related
Categories
Questions about Create Agent
What does Create Agent do?
创建新的 OpenClaw Agent 及其 workspace。包含四个阶段：信息收集、 workspace 构造、系统注册、重启验证。. Create Agent is an agent skill from LeoYeAI/openclaw-master-skills.
When should I use Create Agent?
Create Agent fits situations like: tasks that involve Building AI agents.
How do I install Create Agent in Claude Code?
Run `npx skills add LeoYeAI/openclaw-master-skills --skill create-agent -a claude-code`. Or copy the skill folder (skills/openclaw-create-agent in LeoYeAI/openclaw-master-skills) into .claude/skills/create-agent in your project. Claude Code loads it when a task matches its description.
How do I install Create Agent in Codex?
Run `npx skills add LeoYeAI/openclaw-master-skills --skill create-agent -a codex`. Or copy the skill folder (skills/openclaw-create-agent in LeoYeAI/openclaw-master-skills) into .agents/skills/create-agent in your project. Codex loads it when a task matches its description.
Can I use Create Agent in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add LeoYeAI/openclaw-master-skills --skill create-agent -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/create-agent, .gemini/skills/create-agent, .github/skills/create-agent and .opencode/skills/create-agent in your project.
What does Create Agent need to run?
Going by SKILL.md and its folder, Create Agent needs a shell and Python for the scripts in its folder and the command-line tools its instructions call (bash and python3). Our summary lists: Python 3; A Bash shell.
Does Create Agent access the network?
SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.
Is Create Agent safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.
What licence does Create Agent use?
Create Agent is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Create Agent use?
About 2.4k tokens (SKILL.md is roughly 9.5k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 6.8k tokens, read only when the agent opens those files.
What are the alternatives to Create Agent?
Skills that share tags, products or a category with Create Agent: Agent Builder (shareAI-lab/learn-claude-code, 78k stars), Azure AI Projects Python SDK (microsoft/skills, 3.1k stars), Paperclip Create Agent (paperclipai/paperclip, 98k stars) and Senior Prompt Engineer (maslennikov-ig/claude-code-orchestrator-kit, 259 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Create Agent?
LeoYeAI (a GitHub user) maintains it in LeoYeAI/openclaw-master-skills, which has 2,158 GitHub stars. The repository holds 1,215 skills in this directory. The repository was last updated on July 20, 2026.
Source: LeoYeAI/openclaw-master-skills on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.