Agency Agents Zh

193 个即插即用的 AI 专家角色库 - 覆盖工程、设计、营销、产品、游戏、安全、金融等 18 个部门。支持多智能体协作工作流。

Security Scan

The skill claims a large on-disk library of 193 agent roles located at /root/.openclaw/agency-agents-zh/, but the package contains only SKILL.md and no install or role files—this mismatch and the explicit instructions to read root-path files are inconsistent and worth caution.

Purpose & Capability

The skill's stated purpose is a local role library, yet the package includes only SKILL.md and no role files or install steps. The SKILL.md asserts the project is cloned at /root/.openclaw/agency-agents-zh/, which is not reflected in the registry metadata—this discrepancy suggests either missing install steps or misleading claims.

Instruction Scope

Runtime instructions explicitly reference reading files under /root/.openclaw/agency-agents-zh/ (e.g., cat AGENT-LIST.md and individual role .md files). That directs the agent to access local filesystem paths not declared in the manifest; while reading those specific files could be legitimate for this skill, the instructions give the agent unmediated access to a root-owned path and rely on unspecified preexisting files.

Install Mechanism

No install spec is provided (lowest install risk). However, SKILL.md claims the project is already cloned into a specific path. The absence of an install mechanism plus the assertion of pre-cloned files is inconsistent and should be clarified.

Credentials

The skill requests no environment variables or credentials (appropriate), but it nonetheless expects access to local files under /root/.openclaw/. The manifest declares no required config paths while the instructions rely on a specific root-path dataset—this mismatch increases risk because the skill implicitly requires filesystem access that wasn't declared.

Persistence & Privilege

The skill does not request always: true and is user-invocable with normal autonomous invocation settings. It does not declare any special persistent privileges.

What to consider before installing

This skill is suspicious because its SKILL.md assumes a large repository exists at /root/.openclaw/agency-agents-zh/ but the package contains only the SKILL.md and no install steps. Before enabling: (1) Ask the publisher/source for the repository or an install mechanism; verify the homepage or origin. (2) Locally inspect /root/.openclaw/agency-agents-zh/ (ls -la and open AGENT-LIST.md) to confirm the role files actually exist and review them for unexpected content. (3) If you must use it, run it in a restricted/sandboxed environment and avoid granting broad filesystem or network access until you verify contents. (4) Prefer skills that include their role files or a clear, trustworthy install step and a known source. If the author can provide the missing files or a transparent install process, that would materially increase confidence.

agency-agents-zh 专家角色库

概述

这是 agency-agents-zh 项目的 OpenClaw 集成技能。提供了 193 个专业 AI 专家角色，覆盖 18 个部门，包括 46 个中国市场原创智能体。

安装位置

项目已克隆到：/root/.openclaw/agency-agents-zh/

主要目录结构

agency-agents-zh/
├── academic/          # 学术领域
├── design/           # 设计领域
├── engineering/      # 工程领域（最大）
├── finance/          # 金融领域
├── game-development/ # 游戏开发
├── marketing/        # 市场营销
├── product/          # 产品管理
└── ...

如何使用

方法1: 直接调用特定专家

当需要特定领域的专业知识时，可以：

1. 指定角色名称：如"安全工程师"、"后端架构师"、"UI设计师"
2. 系统会自动加载对应的专家角色定义
3. AI 将按照该专家的思维模式和工作流程执行任务

方法2: 与 workflow-runner 配合使用

对于复杂任务，使用多智能体协作：

1. 创建 YAML 工作流文件
2. 使用 workflow-runner 技能执行
3. 多个专家角色自动协作完成任务

核心专家角色示例

工程领域

• 安全工程师 - OWASP Top 10, 威胁建模, 安全代码审查
• 后端架构师 - 系统设计, 微服务, 数据库优化
• 前端开发者 - 现代前端框架, 性能优化, 用户体验
• DevOps 自动化师 - CI/CD, 容器化, 云原生
• 数据工程师 - 数据管道, ETL, 大数据处理

设计领域

• UI 设计师 - 界面设计, 用户体验, 设计系统
• UX 研究员 - 用户研究, 可用性测试, 用户旅程
• 视觉故事讲述者 - 数据可视化, 信息设计

中国市场原创

• 小红书运营专家 - 种草笔记策略, 达人合作
• 抖音内容策略师 - 短视频内容, 算法优化
• 微信小程序开发者 - 小程序开发, 微信生态
• 飞书集成开发者 - 飞书应用开发, 机器人集成
• 钉钉集成开发者 - 钉钉工作台, 智能助手

工作流程

1. 识别需求 - 确定需要哪种专家
2. 加载角色 - 从 agency-agents-zh 加载对应的角色定义
3. 执行任务 - AI 按照专家的工作流程执行
4. 交付成果 - 输出符合专业标准的成果

与现有技能的整合

• workflow-runner - 用于多智能体协作工作流
• superpowers-writing-plans - 增强规划能力
• security-auditor - 与安全工程师角色互补

注意事项

1. 角色定义文件在 /root/.openclaw/agency-agents-zh/ 目录
2. 每个角色都有完整的专业流程和交付物标准
3. 支持多角色协作完成复杂任务
4. 中国市场角色针对中国本土需求优化

快速开始

# 查看所有可用的专家角色
cat /root/.openclaw/agency-agents-zh/AGENT-LIST.md

# 查看特定角色定义
cat /root/.openclaw/agency-agents-zh/engineering/engineering-security-engineer.md

现在你可以说："作为安全工程师，请审查这段代码" 或 "我需要后端架构师帮助设计系统"。