📦 Mcp Server Scanner — 漏洞扫描器
v1.0.0和 assesses MCP servers 用于 vulnerabilities, insecure configs, 数据 exposure, 和 compliance 使用 SOC 2, GDPR, 和 ISO 27001 standards.
0· 102·0 当前·0 累计
by @engsathiago 安全扫描
OpenClaw
可疑
medium confidenceThe skill claims to perform broad environment discovery and secret-finding but provides no install, no declared required credentials, and only vague instructions — that mismatch is concerning.
评估建议
This skill's goals (discovering MCP servers, finding hardcoded secrets, scanning configs and network) imply access to sensitive system files and network traffic, but the package declares no required credentials, binaries, or file paths and gives only vague instructions. Before installing or invoking it: 1) Ask the author for source code or a trustworthy homepage and for an explicit list of files/paths, network ranges, and credentials the skill will access. 2) Require the skill to declare require...详细分析 ▾
⚠ 用途与能力
The SKILL.md describes discovery (finding MCP servers and agent configs), secret detection, network/TLS checks, and compliance mapping — tasks that normally require access to network scanning tools, host/config paths, or API credentials. Yet the skill declares no required binaries, no config paths, and no environment variables. That mismatch (claiming intrusive capabilities but requesting no explicit access) is unexplained and disproportionate.
⚠ 指令范围
The runtime instructions are high-level and open-ended: they tell the agent to 'run discovery and security scan' but provide no concrete, scoped commands or limits. Because the doc grants broad authority implicitly, it could lead the agent to read arbitrary config files, network endpoints, or secrets unless constrained. There are no explicit allowed paths, endpoints, or safeguards.
✓ 安装机制
No install spec and no code files beyond a minimal package.json — this is instruction-only, so nothing will be downloaded or written to disk by the skill itself. That lowers direct supply-chain risk.
⚠ 凭证需求
The skill requests no environment variables or credentials, yet its stated behavior (detecting hardcoded secrets, mapping agent configs, assessing retention and PII exposure) would normally require access to sensitive data and possibly service credentials. The lack of declared required credentials or explicit data sources is disproportionate and unclear.
✓ 持久化与权限
The skill is not always-enabled and does not request persistent system changes. It can be invoked by the agent (normal default). While autonomous invocation is allowed by default, this alone is not flagged; however, autonomous runs combined with the above ambiguities increase the risk surface.
安全有层次，运行前请审查代码。
运行时依赖
无特殊依赖
安装命令
点击复制官方npx clawhub@latest install mcp-server-scanner
镜像加速npx clawhub@latest install mcp-server-scanner --registry https://cn.longxiaskill.com