← 返回 Skills 市场
6085
总下载
25
收藏
48
当前安装
3
版本数
在 OpenClaw 中安装
/install multi-agent-cn
功能描述
通用多Agent调度系统（中文版）：将主Agent变为纯调度员，所有任务通过 sessions_spawn 委派给5个持久化子Agent。支持轮询调度、先回复再派遣协议、 sessionKey固定复用。用户可自定义调度员角色和子Agent名称/人设。
安全使用建议
Install only if you are comfortable with task details being copied into reusable sub-agent sessions. Avoid sending secrets, account data, production credentials, or unrelated private context through this dispatcher unless you add your own rules for confirmation, project separation, and session reset.
功能分析
Type: OpenClaw Skill
Name: multi-agent-cn
Version: 1.2.0
The skill defines a multi-agent dispatcher. The `SKILL.md` explicitly instructs the main agent (dispatcher) *not* to perform any direct execution (`exec`), file read/write, or search operations. Its role is strictly to delegate tasks to sub-agents via `sessions_spawn`. There are no instructions for data exfiltration, malicious execution, persistence, or prompt injection designed to harm the user or system. The instructions are clear, transparent, and focus on managing task delegation within the OpenClaw framework, even including explicit prohibitions against risky behaviors for the dispatcher agent.
能力评估
Purpose & Capability
The stated purpose and artifacts align: it is a Chinese-language generic dispatcher that tells the main agent to delegate work through sessions_spawn to five named sub-agents.
Instruction Scope
The skill is intentionally generic, says all real work must be delegated, and supports parallel dispatch of multiple tasks, but does not define task categories, sensitive-task exclusions, or confirmation boundaries.
Install Mechanism
The artifact contains only SKILL.md and skill.json, with no executable scripts, package install step, binary requirement, credential setup, or hidden payload.
Credentials
The main dispatcher is explicitly forbidden from direct exec, file read/write, search, and message-tool use; actual tool use is delegated to spawned sub-agents according to the task.
Persistence & Privilege
The artifact requires fixed sessionKey reuse for alpha through echo to preserve sub-agent memory, but does not define project separation, session reset, retention limits, or safeguards for secrets and private context.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install multi-agent-cn - 安装完成后，直接呼叫该 Skill 的名称或使用
/multi-agent-cn触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.2.0
Fix: 强化 sessionKey 必传说明，加正确/错误示例对比，解决 session 不复用一直新增的问题
v1.1.0
新增多任务拆解+并行派遣：用户一句话多个任务时，自动拆解并同时派遣多个子Agent执行
v1.0.0
Initial release: 通用多Agent调度系统中文版 — 5子Agent轮询调度，先回复再派遣，sessionKey固定复用，无主题绑定，用户可自定义角色和名称
元数据
常见问题
Multi-Agent CN 是什么？
通用多Agent调度系统（中文版）：将主Agent变为纯调度员，所有任务通过 sessions_spawn 委派给5个持久化子Agent。支持轮询调度、先回复再派遣协议、 sessionKey固定复用。用户可自定义调度员角色和子Agent名称/人设。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 6085 次。
如何安装 Multi-Agent CN？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install multi-agent-cn」即可一键安装，无需额外配置。
Multi-Agent CN 是免费的吗？
是的，Multi-Agent CN 完全免费（开源免费），可自由下载、安装和使用。
Multi-Agent CN 支持哪些平台？
Multi-Agent CN 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Multi-Agent CN？
由 CLOUD BOY（@be1human）开发并维护，当前版本 v1.2.0。
推荐 Skills