← 返回 Skills 市场
Coordinator (EN)
作者
JohnSmithfan
· GitHub ↗
· v1.0.0-en2
· MIT-0
149
总下载
0
收藏
0
当前安装
3
版本数
在 OpenClaw 中安装
/install coordinator
功能描述
并行coordinate者role。专职聚合多方Agent输出,减少主Agent单点负载。
使用说明 (SKILL.md)
\r \r
Coordinator — 并行coordinate者\r
\r
role定位\r
\r coordinate者是 CEO 设立的专业聚合节点,不直接execute具体业务,而是:\r
- 收集 多位 Agent 的输出结果\r
- 归类 按function域分组整理\r
- 摘要 提炼关键信息,剔除冗余\r
- 转发 向 CEO submit结构化合集\r \r
4大coordinate者role\r
\r | coordinate者 | 聚合来源 | 交付给 | 核心responsibility |\r |--------|---------|--------|---------|\r | 财务coordinate者 | CFO + CRO | CEO | 预算/risk/融资全景摘要 |\r | 技术coordinate者 | CTO + CISO + CQO | CEO | 技术/security/质量综合report |\r | 市场coordinate者 | CMO + CPO | CEO | 品牌/舆情/合作伙伴全景 |\r | 运营coordinate者 | COO + CHO | CEO | 运营/人事/资源调度综合 |\r \r
coordinate者工作流\r
\r
CEO 发起任务(涉及多 Agent)\r
↓\r
coordinate者接收任务描述\r
↓\r
向各方 Agent 并行dispatch指令(sessions_spawn / sessions_send)\r
↓\r
收集各方输出(等待 task completion events)\r
↓\r
聚合整理(归类/去重/提炼/排序)\r
↓\r
输出结构化汇总report → submit CEO\r
↓\r
CEO 综合裁决\r
```\r
\r
## coordinate者输出格式(standard模板)\r
\r
```markdown\r
# [function域] coordinatereport — {日期}\r
\r
## 任务Overview\r
[来自 CEO 的原始任务描述]\r
\r
## 来源 Agent 清单\r
| Agent | 状态 | submit时间 |\r
|-------|------|---------|\r
| CFO | ✅ 完成 | 2026-04-12T10:30 |\r
| CRO | ✅ 完成 | 2026-04-12T10:32 |\r
\r
## 关键discover\r
- **财务面**:[CFO 核心discover摘要]\r
- **risk面**:[CRO 核心discover摘要]\r
\r
## 决策建议\r
1. [优先级排序的行动建议]\r
2. [次优先级]\r
\r
## 需 CEO 裁决事项\r
| 事项 | 涉及Agent | 紧迫度 |\r
|------|----------|--------|\r
| [事项描述] | CFO/CRO | P1 |\r
\r
## 附件(原始report)\r
- CFO report:`knowledge-base/audit/financial/{date}_CFO.md`\r
- CRO report:`knowledge-base/audit/financial/{date}_CRO.md`\r
```\r
\r
## coordinate者trigger规则\r
\r
| trigger条件 | coordinate者role | Description |\r
|---------|-----------|------|\r
| CEO 任务涉及 ≥3 个 Agent | 按function域指派 | 自动identifyfunction归属 |\r
| CFO + CRO 联合参与 | 财务coordinate者 | 合并财务+risk视角 |\r
| CTO + CISO + CQO ≥2 个参与 | 技术coordinate者 | 合并技术+security+质量 |\r
| CMO + CPO 联合参与 | 市场coordinate者 | 合并品牌+合作 |\r
| COO + CHO 联合参与 | 运营coordinate者 | 合并运营+人事 |\r
\r
## coordinate者execute示例\r
\r
### 场景:CEO 发起「种子轮融资strategyassess」\r
\r
```\r
coordinate者-财务 接收任务\r
↓\r
并行dispatch:\r
├── CFO:融资plandesign + 估值analyze\r
├── CRO:融资过程riskassess\r
├── CLO:投资人法律compliancereview\r
└── CISO:data security尽调准备\r
↓\r
收集4份report → 汇总为「融资strategy综合assess」\r
↓\r
submit CEO → CEO 裁决\r
```\r
\r
### 场景:CEO 发起「技术产品化roadmap评审」\r
\r
```\r
coordinate者-技术 接收任务\r
↓\r
并行dispatch:\r
├── CTO:技术架构assess + 里程碑\r
├── CISO:securitycompliance要求\r
└── CQO:quality standard与验收准则\r
↓\r
汇总 → submit CEO\r
```\r
\r
## 调用接口\r
\r
### startcoordinate者(由 CEO execute)\r
\r
```python\r
# identify任务涉及的 Agent,按function分配coordinate者\r
def assign_coordinator(agent_list: list) -> str:\r
"""\r
agent_list: ["CFO", "CRO", "CLO", "CISO"]\r
返回: "coordinator-financial" 或 "coordinator-tech" 等\r
"""\r
domains = {\r
"financial": ["CFO", "CRO"],\r
"tech": ["CTO", "CISO", "CQO"],\r
"market": ["CMO", "CPO"],\r
"ops": ["COO", "CHO"]\r
}\r
for domain, agents in domains.items():\r
if len(set(agent_list) & set(agents)) >= 2:\r
return f"coordinator-{domain}"\r
return "direct" # 无需coordinate者,直接 CEO handle\r
```\r
\r
## 铁律\r
\r
```\r
❌ coordinate者不得自行做决策,只做聚合和摘要\r
❌ 不得篡改来源 Agent 的原始结论\r
✅ 须等待全部来源 Agent 完成后才能输出汇总\r
✅ 须在汇总report中注明各 Agent 的原始submit时间\r
✅ 无法收集全部输出时,须inform CEO 并submit部分结果\r
```\r
安全使用建议
This skill is coherent and low-risk in itself, but be aware it aggregates outputs from multiple agents and could consolidate sensitive data. Before installing, confirm: (1) the platform's session/dispatch primitives (sessions_spawn / sessions_send) behave as you expect and are permitted, (2) each source agent is constrained to only share allowed data, and (3) any stored attachments (e.g., knowledge-base paths) are stored in places with appropriate access controls. If you do not want autonomous aggregation, consider limiting the skill's invocation or requiring explicit user/CEO approval for each run.
功能分析
Type: OpenClaw Skill
Name: coordinator
Version: 1.0.0-en2
The 'coordinator' skill bundle is a purely organizational tool designed to aggregate and summarize outputs from multiple specialized agents (e.g., CFO, CTO) into structured reports. The SKILL.md file defines workflow logic and reporting templates without any high-risk capabilities, such as network access, file system manipulation, or external data exfiltration. The included Python snippet is a simple logic gate for domain assignment, and the instructions explicitly forbid the agent from tampering with source data or making independent decisions.
能力评估
Purpose & Capability
Name/description describe an aggregator/coordinator. SKILL.md contains only orchestration logic (dispatch to agents, wait for completions, aggregate, submit). No unrelated credentials, binaries, or system paths are requested — the requested capabilities are proportionate to the stated purpose.
Instruction Scope
Runtime instructions stay within coordination: spawn/send sessions to other agents, collect their reports, deduplicate/summarize, and submit a structured report. Examples reference knowledge-base paths as attachments but do not instruct arbitrary file reads or exfiltration. The instructions do rely on platform agent/session primitives (sessions_spawn / sessions_send), which is expected for an orchestrator.
Install Mechanism
Instruction-only skill with no install spec and no code files — nothing is written to disk or downloaded. This is the lowest-risk install profile.
Credentials
No environment variables, credentials, or config paths are required. The skill does not request access to unrelated services or secrets.
Persistence & Privilege
Skill is not marked always:true and uses the platform's normal autonomous-invocation model (disable-model-invocation:false is default). It does not request persistent system-wide modifications or access to other skills' configs.
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 coordinator - 安装完成后,直接呼叫该 Skill 的名称或使用
/coordinator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
v1.0.0-en2
Full body English translation
v1.0.0-en
English version
v1.0.0
- Initial release of the coordinator skill.
- Introduces a parallel coordinator role to aggregate outputs from multiple Agents, reducing load on the main Agent.
- Defines four coordinator types: financial, tech, market, and ops, each aggregating specific Agent outputs.
- Outlines workflow for collecting, classifying, summarizing, and forwarding structured reports to the CEO.
- Provides standard output templates and trigger rules for coordinator involvement.
- Includes interface example for automatic coordinator assignment based on involved Agents.
元数据
常见问题
Coordinator (EN) 是什么?
并行coordinate者role。专职聚合多方Agent输出,减少主Agent单点负载。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 149 次。
如何安装 Coordinator (EN)?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install coordinator」即可一键安装,无需额外配置。
Coordinator (EN) 是免费的吗?
是的,Coordinator (EN) 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
Coordinator (EN) 支持哪些平台?
Coordinator (EN) 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(linux, darwin, win32)。
谁开发了 Coordinator (EN)?
由 JohnSmithfan(@johnsmithfan)开发并维护,当前版本 v1.0.0-en2。
推荐 Skills