中文 EN

扫描报告

扫描新文件

⚠ 低风险 — 风险评分 15/100
上次扫描：1 天前 重新扫描
15 /100
scout-apm
Scout APM integration — query traces, endpoints, errors, N+1 queries, and deployments via Membrane CLI
Scout APM integration skill using the Membrane CLI with appropriate declared permissions and no malicious behavior detected.
技能名称scout-apm
分析耗时32.8s
引擎pi
✓
可以安装
Approve for use. Consider pinning the npm CLI version for supply chain hygiene.

安全发现 2 项

严重性 安全发现 位置
低危
Unpinned npm global package 供应链
The skill installs @membranehq/cli without a pinned version, using 'npm install -g @membranehq/cli'. This allows the package maintainer to publish a new version that could change behavior or be compromised.
npm install -g @membranehq/cli
→ Pin to a specific version: npm install -g @membranehq/[email protected]
 SKILL.md:31
低危
Implied but not declared network:READ scope 文档欺骗
The skill proxies requests to Scout APM API via membrane but does not explicitly declare network:READ in its allowed-tools mapping, though it is implicitly required.
No allowed-tools or capabilities declaration present in frontmatter
→ Add an allowed-tools mapping to SKILL.md frontmatter that explicitly declares network:READ for the membrane request proxy functionality
 SKILL.md:1
资源类型声明权限推断权限状态证据
网络访问 READ READ ✓ 一致 SKILL.md:47-64 — membrane request/connect operations for Scout APM API
命令执行 WRITE WRITE ✓ 一致 SKILL.md:31 — npm install -g @membranehq/cli; SKILL.md:35 — membrane login/conne…
文件系统 NONE NONE — No filesystem operations found
环境变量 NONE NONE — No environment variable access detected
技能调用 NONE NONE — No nested skill invocations
剪贴板 NONE NONE — No clipboard access
浏览器 NONE NONE — No browser automation
数据库 NONE NONE — No database access
2 项发现
🔗
中危 外部 URL 外部 URL
https://getmembrane.com
SKILL.md:7
🔗
中危 外部 URL 外部 URL
https://docs.scoutapm.com/
SKILL.md:19

目录结构

1 文件 · 4.4 KB · 127 行
Markdown 1f · 127L
└─ 📝 SKILL.md Markdown 127L · 4.4 KB

依赖分析 1 项

包名版本来源已知漏洞备注
@membranehq/cli * npm 否 No version pinned; installed globally via npm install -g

安全亮点

✓ No credential theft — skill explicitly delegates auth to Membrane CLI and states never to ask for API keys
✓ No sensitive path access — no reads of ~/.ssh, ~/.aws, .env, or similar
✓ No obfuscation — all code is plain text, no base64 or eval
✓ No data exfiltration — all network calls are to documented external services (getmembrane.com, Scout APM API)
✓ No reverse shell or C2 infrastructure
✓ No hidden HTML comments or embedded payloads
✓ Clear documentation matching actual behavior — no doc-to-code mismatch
✓ Credential lifecycle handled server-side by Membrane, not stored locally