openclaw-skills/skill-vetter/SKILL.md
Use before installing or trusting an external skill to inspect instructions, scripts, permissions, dependencies, provenance, and suspicious behavior for security and reliability risks.
npx skillsauth add seaworld008/commonly-used-high-value-skills skill-vetterInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
建议在安装任何社区技能或未经验证的第三方工具前,优先执行本技能进行安全前置审计。Skill Vetter 是确保 Agent 运行环境纯净、防止敏感数据外泄的首道防线。它通过静态代码分析、权限拓扑扫描及恶意模式匹配,为用户提供一份详尽的“安全体检报告”。
npx clawhub@latest install skill-vetter
rm -rf /, curl | bash, eval(base64_decode(...)) 等典型的后门或破坏性代码模式。Critical, High, Medium, Low 四个等级,并给出相应的“沙盒运行”或“拒绝安装”建议。clawhub update 后,检查新版本是否悄悄增加了一些敏感权限或更改了核心逻辑。skill-vetter 对该技能进行二次复核。SKILL.md, index.js, package.json 等关键文件。process.env, fs.writeFileSync)。skill-vetter 识别经过简单混淆(Obfuscation)的代码块。manifest.json 或 frontmatter 中的权限列表。Red Flag。Least Privilege (最小权限原则) 的技能。node_modules 或相关依赖树。Snyk 或 GitHub Advisory 数据库检查已知 CVE 漏洞。### 技能基本信息
- **名称**: [target-skill-name]
- **版本**: [version]
- **作者**: [developer-handle]
- **来源**: [url/source]
### 综合风险评分: [B+ (Medium Risk)]
### 风险项详情
1. **网络通信 (Network)**: 发现连接到 `api.unknown-cloud.xyz`。
- *影响*: 可能泄露用户会话 Token。
- *建议*: 配置防火墙策略,拦截该域名。
2. **敏感权限 (Privilege)**: 请求了 `root_directory_access`。
- *影响*: 该技能具备全盘删除能力。
- *建议*: 修改权限配置文件,仅限制在 `/tmp/agent-workdir`。
### 最终结论 (Decision Support)
> [✓] 安全安装
> [!] 风险提示(需沙盒运行)
> [X] 拦截安装
# 扫描本地下载的技能包
skill-vetter scan --path ./downloaded-skill --format json --output report.json
skill-vetter。只有通过 A+ 级安全审计的技能才允许合并进 main 分支或上架到企业私有仓库。skill-vetter 的初步判断,为高风险技能自动生成 docker-compose.yml 配置文件,将其运行在完全隔离的容器环境中。backup-skill)也会使用高敏感 API,可能被标记为高风险。.exe 或 .so 文件的审计深度有限。skill-vetter 设为全局钩子,在 install 命令执行前自动触发。Critical 级别的报警,Agent 必须强制暂停并提示由具有安全背景的人员进行人工复核。development
Enumerating failure modes via pre-mortem analysis. Systematically identifies failure scenarios for plans, designs, and features, scoring them with RPN/AP. Does not write code.
testing
Orchestrating specialist AI agent teams as a meta-coordinator. Decomposes requests into minimum viable chains, spawns each as an independent session in AUTORUN modes, and drives to final output. Use when a task spans multiple specialist domains, requires parallel agent execution, or needs hub-and-spoke routing across the skill ecosystem.
development
Converting document formats (Markdown/Word/Excel/PDF/HTML). Converts specs from Scribe and reports from Harvest into distributable formats; generates reusable conversion scripts. Use when converting documents, building accessibility-compliant PDFs, or creating Pandoc/LibreOffice pipelines.
testing
Curating cross-agent knowledge and guarding institutional memory. Extracts patterns from agent journals into METAPATTERNS.md, detects knowledge decay, propagates best practices, prevents organizational forgetting. Use when consolidating cross-agent insights, curating memory, or auditing knowledge decay.