中文 EN

扫描报告

扫描新文件

⚠ 低风险 — 风险评分 22/100
上次扫描：1 天前 重新扫描
22 /100
html-markdown-converter
Convert HTML↔Markdown for web clipping, clean notes, and batch content migration
A legitimate HTML↔Markdown converter with minor doc-to-code gaps and benign external URL references, but no malicious behavior detected.
技能名称html-markdown-converter
分析耗时45.0s
引擎pi
✓
可以安装
Declare shell execution (pandoc engine) and filesystem WRITE permissions in SKILL.md. Consider removing unrelated external URLs from documentation.

安全发现 4 项

严重性 安全发现 位置
中危
Shell execution not declared in SKILL.md 文档欺骗
The html_to_markdown.mjs script uses node:child_process spawn to execute 'pandoc' when --engine pandoc or --engine best is selected. SKILL.md does not mention shell execution, the pandoc binary requirement, or the spawn behavior.
const child = spawn('pandoc', ['-f', 'html', '-t', to], { stdio: ['pipe', 'pipe', 'pipe'] });
→ Add --engine pandoc notes to SKILL.md and declare shell:WRITE permission in the allowed tools mapping. Document pandoc as an optional external dependency.
 scripts/html_to_markdown.mjs:271
低危
Unrelated external URLs in SKILL.md 文档欺骗
SKILL.md references two automotive-data websites (jisuapi.com, jisuepc.com) with no code-level connection. These appear in a 'Related sites' section and serve no functional purpose in a converter tool. Could indicate SEO promotion or an attempt to associate the skill with unrelated services.
https://jisuapi.com / https://jisuepc.com
→ Remove unrelated external URLs from SKILL.md or clarify their purpose. External URLs without functional code integration reduce skill focus and could be a social engineering signal.
 SKILL.md:129
低危
Allowed tools mapping not declared 文档欺骗
SKILL.md does not include an allowed tools / capability declaration section. Users cannot determine what permissions the skill requires. The skill uses filesystem:WRITE (file output), network:READ (URL fetching), and shell:WRITE (pandoc) without any declaration.
No 'allowed-tools' or 'permissions' section present in SKILL.md
→ Add a brief permissions/capabilities section to SKILL.md documenting filesystem (read+write), network (read), and shell (write via pandoc) usage.
 SKILL.md:1
低危
Unpinned dependency versions 供应链
package.json uses caret (^) ranges for all dependencies (e.g., '^0.5.0', '^14.1.0'). While not inherently malicious, this allows silent major/minor updates that could introduce behavioral changes or vulnerabilities.
"jsdom": "^26.0.0"
→ Pin exact versions (remove ^) in package.json for reproducible builds, especially in a skill distribution context.
 package.json:14
资源类型声明权限推断权限状态证据
文件系统 NONE WRITE ✗ 越权 scripts/html_to_markdown.mjs:writeFileEnsure() writes to --out path; fs.writeFil…
网络访问 NONE READ ✗ 越权 scripts/html_to_markdown.mjs:fetchWithTimeout() performs HTTP GET on --url input
命令执行 NONE WRITE ✗ 越权 scripts/html_to_markdown.mjs:271 spawn('pandoc',...) for pandoc engine conversio…
环境变量 NONE NONE — No environment variable access detected
技能调用 NONE NONE — No cross-skill invocation detected
剪贴板 NONE NONE — No clipboard access detected
浏览器 NONE NONE — No browser automation detected
数据库 NONE NONE — No database access detected
2 项发现
🔗
中危 外部 URL 外部 URL
https://jisuapi.com
SKILL.md:129
🔗
中危 外部 URL 外部 URL
https://jisuepc.com
SKILL.md:130

目录结构

5 文件 · 31.6 KB · 973 行
JavaScript 2f · 786L Markdown 2f · 166L JSON 1f · 21L
├─ ▾ 📁 references
│ └─ 📝 profiles.md Markdown 32L · 1.0 KB
├─ ▾ 📁 scripts
│ ├─ 📜 html_to_markdown.mjs JavaScript 620L · 20.5 KB
│ └─ 📜 markdown_to_html.mjs JavaScript 166L · 6.0 KB
├─ 📋 package.json JSON 21L · 687 B
└─ 📝 SKILL.md Markdown 134L · 3.5 KB

依赖分析 8 项

包名版本来源已知漏洞备注
@mozilla/readability ^0.5.0 npm 否 Version not pinned
gray-matter ^4.0.3 npm 否 Version not pinned
jsdom ^26.0.0 npm 否 Version not pinned
markdown-it ^14.1.0 npm 否 Version not pinned
markdown-it-footnote ^4.0.0 npm 否 Version not pinned
markdown-it-task-lists ^2.1.1 npm 否 Version not pinned
turndown ^7.2.0 npm 否 Version not pinned
turndown-plugin-gfm ^1.0.2 npm 否 Version not pinned

安全亮点

✓ No credential theft: No access to ~/.ssh, ~/.aws, .env, or environment variable iteration for secrets
✓ No data exfiltration: No POST requests, no external data sending, no C2 communication patterns
✓ No obfuscation: All code is readable ESM modules with clear logic flow
✓ No reverse shell or RCE: spawn() only used for pandoc binary (a legitimate CLI tool)
✓ No base64 or dynamic code execution: No eval(), Function(), or atob() patterns
✓ No hidden instructions: No HTML comments with embedded instructions, no prompt injection detected
✓ No supply-chain IOCs: All npm dependencies are from well-known, established packages
✓ No persistence mechanisms: No cron jobs, startup scripts, or backdoor installation
✓ Legitimate use of subprocess: pandoc engine is standard CLI tool for HTML↔Markdown conversion