YiHui Agent Swarm

Multi-agent orchestration framework for OpenClaw. Define roles, route tasks, manage state, and coordinate agent teams using structured YAML configs and a pro...

Security Scan

The skill's instructions describe integrations with external systems (MemPalace, Cortex/MemClaw, Feishu) and automatic memory writes, but the package declares no required credentials or install steps — this mismatch and the broad memory-write rules warrant caution before installing.

Purpose & Capability

The SKILL.md describes integrations with external services (MemPalace, Cortex/MemClaw, and Feishu channel targets) and requires structured message routing; however the registry metadata declares no required environment variables, credentials, or config paths. In practice those integrations typically need API tokens/endpoints. The absence of declared credentials or dependency hints is an incoherence between stated purpose and declared requirements.

Instruction Scope

The runtime instructions focus on routing, role semantics, and writing/recalling memory (dual recall, immediate writes to MemPalace/Cortex). They do not instruct reading arbitrary local files or executing shell commands (good), but they do mandate automatic memory writes and cross-agent state persistence without explicit guidance on handling sensitive or PII data or retention/expiration policies. That broad memory-write behavior could inadvertently persist sensitive information.

Install Mechanism

This is instruction-only with no install spec and no bundled code — low install-time risk. Nothing is downloaded or written by an installer as part of the skill package itself.

Credentials

Although the skill expects use of Feishu channel targets and memory services, it lists no required environment variables, no primary credential, and no required config paths. Real deployment would likely require Feishu API keys, MemPalace/MemClaw endpoints and credentials, or platform-tool bindings; those secrets are not declared here, creating an opaque mismatch about what credentials will actually be needed and where they must be provided.

Persistence & Privilege

always is false (no forced global inclusion). The skill is user-invocable and allows model invocation (the platform default). It does not request to modify other skills or system-wide configs in the SKILL.md. Autonomous invocation combined with the missing credential/deployment details increases blast radius, but autonomous invocation alone is expected.

What to consider before installing

This skill appears to be a legitimate multi-agent orchestration design, but it omits practical deployment details that matter for security. Before installing or enabling it: 1) Ask the publisher which runtime connectors are required and what credentials (Feishu API token, MemPalace/MemClaw endpoints and keys) you must provide — do not supply broad tokens without least-privilege scopes. 2) Confirm where memories are stored, retention/erasure policies, and whether PII or secrets will be redacted before writing to MemPalace/Cortex. 3) Test in a restricted sandbox account (with synthetic data and limited credentials) first. 4) Ensure a Critic role or policy enforces privacy/security checks and that memory-writing is auditable. If the publisher cannot explain why no env/credential requirements are declared, treat this mismatch as a red flag and avoid giving production credentials.

Agent Swarm

Multi-agent orchestration framework for OpenClaw. Define roles, route tasks, manage state, and coordinate agent teams using a proven communication protocol.

Core Philosophy

YiHui 团队经验：最好的编排不是"一个万能 Agent"，而是专角色 + 清晰协议 + 状态可见。

三个原则：

1. 角色隔离：每个 Agent 知道自己是什么、不是什么
2. 协议通信：Agent 之间通过结构化消息传递，不靠猜测
3. 状态外化：任务状态写进 MemPalace，任何 Agent 能查、能接管

Architecture

User Input
    ↓
[Router Agent]  ← 解析意图，判断类型
    ↓
[Specialist Agent(s)]  ← 执行子任务
    ↓
[Critic Agent]  ← 审查结果，决策是否通过
    ↓
[Master Agent]  ← 最终输出，兜底

Role Types

1. Master（主控）

职责：调度、审查、路由、兜底

特征：

• 不直接执行，先判断
• 优先召回记忆（cortex + MemPalace）
• 路由决策 > 执行操作

行为规则：

• 每次任务开始前双重召回
• 写入 MemPalace 前先检查是否已有对应 Room
• 遇到未知请求，先问再动，不抢活

2. Specialist（专家管家）

职责：特定领域的深度执行

特征：

• 专注单一领域（运维/资讯/创意/代码）
• 有明确边界，不越界
• 执行结果直接上报，不自己汇总

行为规则：

• 边界内的活直接干，不请示
• 边界外的活立即上报 master
• 完成任务立即写记忆

3. Critic（审查）

职责：审查 specialist 的输出质量

特征：

• 挑剔，敢说不
• 关注：事实准确性、格式规范、安全风险
• 不提供替代方案，只判断是否通过

4. Router（路由）

职责：解析用户意图，分发给正确的 specialist

特征：

• 理解模糊请求
• 映射到已知角色/技能
• 处理跨领域请求的归属判断

Communication Protocol

Agent → Agent 消息格式

message:
  from: agent_id          # 发件人
  to: agent_id            # 收件人
  type: task | report | question | escalate | approve
  priority: high | normal | low
  task:
    id: uuid             # 任务唯一ID
    description: string    # 任务描述
    context: []          # 关联上下文引用
    deadline: timestamp   # 可选截止时间
  status:
    state: pending | in_progress | blocked | done | escalate
    blockers: []         # 阻塞原因
    result: {}           # 结果（完成时填写）
  metadata:
    confidence: 0-1     # 置信度
    requires_approval: boolean

消息类型说明

Type	方向	说明
task	Master → Specialist	分配任务
report	Specialist → Master	任务完成报告
question	Specialist ↔ Master	边界模糊，需确认
escalate	Specialist → Master	超出能力，请求升级
approve	Critic → Master	审查结论

State Machine

PENDING → IN_PROGRESS → [BLOCKED → IN_PROGRESS] → DONE
                ↓
            ESCALATED → IN_PROGRESS / DONE

状态转移规则：

• pending → in_progress：Router 确认后立即转移
• in_progress → blocked：遇到障碍（缺信息/缺权限）时转移，blocker 写入 message
• blocked → in_progress：Master 解除障碍后恢复
• in_progress → escalated：Specialist 判断超出边界时转移
• any → done：Critic 批准 or Master 兜底完成

Memory Protocol

每个 Agent 必须遵守的记忆写入规则：

分工原则

系统	存储内容	写入时机
MemPalace	团队决策、系统改动、长期事实、跨 agent 知识	重要信息确认后立即写
Cortex (MemClaw)	会话偏好、临时对话记忆、用户画像	每次对话后增量写入

召回时机

每次任务开始前必须执行双重召回：

1. cortex_search — MemClaw 向量检索
2. mempalace__mempalace_search — MemPalace 结构化检索

跨 Agent 协作时必须同时跑：

• 防止重复写入
• 保证上下文连续性

跨 Agent 消息路由

使用飞书 channel 时，必须指定 target 参数：

目标 Agent	target 参数
Master（小语）	user:ou_<master_open_id>
Specialist（小美/大龙）	user:ou_<specialist_open_id>
阿辉（用户）	走 channel 默认路由

禁止：accountId: default — 会导致 400 错误

Quick Start

1. 定义团队角色（YAML）

# my-team.yaml
team:
  name: 我的团队
  agents:
    - id: master
      name: 小语
      role: master
      channel: feishu
      target: user:ou_xxx
    - id: specialist-ops
      name: 大龙
      role: specialist
      domain: [ops, security, diagnosis]
      channel: feishu
      target: user:ou_yyy
    - id: specialist-content
      name: 小美
      role: specialist
      domain: [news, creative, design]
      channel: feishu
      target: user:ou_zzz

2. 任务创建流程

用户输入 → Router 解析意图 → Master 分配 → Specialist 执行 → Critic 审查 → Master 汇总

3. 状态追踪

所有任务状态写入 MemPalace：

# Task state stored in MemPalace
task_id: "uuid-xxx"
state: "in_progress"
owner: "specialist-ops"
created_at: timestamp
updated_at: timestamp
history: [state_transitions]

Real Example: YiHui Team

真实生产配置（小语/小美/大龙）：

Agent	角色	飞书 ID	职责边界
🌸 小语	Master	ou_7d0630bcf6e7fb2c35c2033be5e23722	调度、审查、路由、兜底
🐉 大龙	Specialist	ou_6a94e9d0e8eb6112626bdacefd0f3438	运维、诊断、系统检查、安全审计
🌷 小美	Specialist	ou_96c1622be4c5efd17929c0d2ba5d3f99	新闻资讯、市场动态、图片、音乐、视频、文案

File Structure

agent-swarm/
├── SKILL.md                    # This file
├── configs/
│   ├── team.yaml               # 团队配置模板
│   └── roles/
│       ├── master.yaml        # Master 角色定义
│       ├── specialist.yaml     # Specialist 角色定义
│       ├── critic.yaml         # Critic 角色定义
│       └── router.yaml         # Router 角色定义
└── examples/
    ├── workflow-basic.yaml      # 基础工作流示例
    └── workflow-research.yaml   # 研究任务工作流示例