Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

teamarchitect-skill

v1.0.0

通过对话收集需求,设计多智能体团队架构,并推荐对应 Agent 角色的专业技能供选择和安装。

0· 107·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for marshong-86/teamarchitect-skill.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "teamarchitect-skill" (marshong-86/teamarchitect-skill) from ClawHub.
Skill page: https://clawhub.ai/marshong-86/teamarchitect-skill
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install teamarchitect-skill

ClawHub CLI

Package manager switcher

npx clawhub@latest install teamarchitect-skill
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
The skill's purpose (design multi-agent team and recommend/install skills) aligns with its behavior: it checks local skills, queries ClawdHub/GitHub, and may create new skills. The meta.json permissions (bash, network, filesystem: skills-directory) are consistent with installation duties. Minor inconsistency: registry summary lists no required config paths, but SKILL.md and README reference ~/.claude/skills/ and meta.json grants filesystem access.
Instruction Scope
SKILL.md stays within stated purpose (collect requirements, map roles to skills, check local installations, fetch from ClawdHub/GitHub, create new skills). It instructs reading the user's skills directory and fetching from external sources — which is expected for an installer but expands the agent's scope to downloading and writing code. Instructions do not specify validation or allowlisting of external sources.
!
Install Mechanism
There is no packaged install spec (instruction-only), but README and SKILL.md explicitly instruct using network actions (git clone, ClawdHub downloads) and meta.json allows bash/network. Downloading and writing external skill code is plausible and not automatically vetted; the install mechanism therefore carries supply‑chain risk because arbitrary repositories or user‑specified sources may be executed/installed.
Credentials
The skill does not request environment variables or secret credentials. That is proportionate to its stated function. However, the skill requests network and shell permissions (via meta.json) which are powerful even without secrets because they allow fetching and executing external code.
!
Persistence & Privilege
always:false (good). But the skill is allowed to read/write the skills directory and run shell/network actions and the platform default permits autonomous invocation. Combined, these privileges could let the agent autonomously download and install additional skills if given consent or if it acts autonomously — increasing the blast radius if a malicious repo is used.
What to consider before installing
This skill appears to do what it says (design teams and recommend/install skills), but it needs shell and network access and will read/write your skills directory to fetch or create skills. Before installing: 1) Prefer not to grant autonomous-installation privileges — require manual confirmation for every install or disable autonomous invocation for this skill. 2) Inspect any recommended skill's source (ClawdHub/GitHub) before approving install; check maintainers and code. 3) Restrict its filesystem scope if possible (limit to a dedicated skills folder). 4) Note the README suggests cloning from a GitHub repo (MarsHong-86) — verify that URL and the repo's contents. 5) If you cannot audit downstream skill code, treat installs as higher risk and avoid enabling automatic batch installs.

Like a lobster shell, security has layers — review code before you run it.

latestvk974sq40nhb34nn0f3ynxyg971838k7d
107downloads
0stars
1versions
Updated 1mo ago
v1.0.0
MIT-0

TeamArchitect - 智能团队架构师

技能描述

通过自然语言对话和多轮询问,深入了解用户需求,设计多智能体团队架构,并根据每个 Agent 角色推荐对应的专业技能供用户选择安装。

核心流程

1. 需求收集阶段

使用引导式对话了解以下信息:

  • 项目类型: 是什么类型的项目/产品
  • 目标用户: 服务对象是谁
  • 技术栈偏好: 现有技术栈或偏好
  • 协作平台: 使用的工具(GitHub、飞书、钉钉等)
  • 团队规模: 预期 Agent 数量
  • 预算限制: 是否有 API 调用成本考虑

2. 架构设计阶段

根据需求智能匹配团队架构,输出:

【团队架构图】
协调 Agent
├── 分析 Agent
├── 执行 Agent
├── 创作 Agent
└── ...

【各 Agent 职责】
- 协调 Agent: 任务分发、上下文管理
- 分析 Agent: 数据收集、信息分析
- ...

3. Agent 角色 - 技能映射库

根据 Agent 角色类型,动态推荐对应技能:

Agent 角色核心技能(主业必需)非核心技能(能力增强)
分析类数据分析、信息抓取、摘要提炼趋势分析、竞品对比
执行类代码编写、文件处理、API 调用批量操作、自动化脚本
审查类代码审查、安全扫描、质量检查性能分析、规范检查
协调类任务分发、上下文管理、进度追踪报告生成、通知提醒
创作类文案撰写、内容优化、排版美化多平台适配、视频脚本
陪伴类对话交流、情感分析、记忆管理个性化推荐、日程提醒
办公类Excel、PPT、Word、PDF 处理邮件管理、会议安排
技术类前后端开发、数据库、运维部署监控告警、CI/CD
研究类深度调研、文献分析知识图谱、引用管理
运营类多平台发布、粉丝互动数据分析、推广策略
设计类UI/UX、视觉设计、图片处理原型设计、素材管理
测试类自动化测试、Bug 检测性能测试、覆盖率分析
学习类知识管理、技能更新记忆整理、经验总结

4. 技能推荐阶段

动态推荐逻辑

对于每个 Agent 角色:
  1. 查找该角色对应的核心技能列表
  2. 检查本地是否已安装
  3. 未安装的标记为 [待安装/待创建]
  4. 汇总生成推荐清单

推荐输出格式

根据你的团队架构,为各 Agent 推荐以下技能:

【分析 Agent】核心技能
  ✓ self-improving - 持续学习改进 [已安装]
  □ data-analyzer - 数据分析 [ClawdHub]
  □ web-scraper - 网页抓取 [待创建]

【办公 Agent】核心技能
  □ excel-handler - Excel 处理 [ClawdHub]
  □ ppt-maker - PPT 制作 [待创建]
  ...

【非核心技能 - 可选增强】
  □ trend-analyzer - 趋势分析 [ClawdHub]
  □ video-scripter - 视频脚本 [待创建]
  ...

---
操作:
1. 安装全部核心技能 [Y/n]
2. 选择安装核心技能 [输入编号]
3. 选择非核心技能 [空格分隔编号]
4. 跳过 [skip]

5. 技能安装策略

对于每个推荐技能:
1. 本地已安装 → 跳过
2. ClawdHub 有现成 → 推荐安装
3. GitHub 有开源 → 推荐下载
4. 都不存在 → 询问是否创建新技能

6. 安装执行阶段

  • 核心技能:用户确认后批量安装/创建
  • 非核心技能:根据用户选择安装
  • 安装失败时提供备选方案
  • 缺失技能:询问是否现场创建

7. 配置优化阶段

  • 检查技能兼容性
  • 提供配置文件调优建议
  • 设置 Agent 间协作规则
  • 创建技能 - 能力映射表供运行时调用

技能来源优先级

  1. 本地已有技能 (~/.claude/skills/)
  2. ClawdHub 技能市场
  3. GitHub 开源技能仓库
  4. 用户指定来源
  5. 现场创建新技能

对话示例

用户:我想做一个自动化代码审查工具

助手:好的,我来帮你设计团队架构。请问:
1. 这个项目主要服务于什么编程语言?
2. 需要集成到哪些平台?(GitHub/GitLab/本地)
3. 预期每天处理多少代码审查请求?

用户:主要是 Python 和 JavaScript,集成 GitHub,每天大约 50 次请求

助手:根据你的需求,推荐以下团队架构:

【团队架构】
├── 代码分析 Agent (负责解析代码结构)
├── 规范审查 Agent (负责检查代码规范)
├── 安全扫描 Agent (负责漏洞检测)
└── 报告生成 Agent (负责生成审查报告)

【为各 Agent 推荐技能】

【分析 Agent】核心技能
  □ code-parser - 代码解析 [ClawdHub]
  □ ast-analyzer - AST 分析 [待创建]

【审查 Agent】核心技能
  ✓ skill-vetter - 技能安全审查 [已安装]
  □ linter-pro - 代码规范检查 [ClawdHub]

【执行 Agent】核心技能
  ✓ debug-pro - 错误调试分析 [已安装]
  □ auto-fixer - 自动修复 [待创建]

是否安装以上核心技能?[Y/n/skip]

用户:Y

助手:正在安装核心技能...
✓ code-parser 安装完成
✓ linter-pro 安装完成
✓ debug-pro 已存在
⚠ auto-fixer 不存在,是否现在创建?[Y/n]

安装命令

/teamarchitect

元数据

  • 版本:1.1.0
  • 作者:用户自定义
  • 适用平台:OpenClaw, Claude Code, 通用 AI 助理

Comments

Loading comments...