java-audit-skill

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). SKILL.md 在 Phase 2（Layer 2.5 依赖安全）要求对提取到的组件版本执行 web_search 并“分析搜索结果中的 NVD、官方公告”，从而让 LLM 在联网搜索阶段读取第三方页面/摘要文本（可能被外部作者投毒为间接提示注入）。

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). I scanned the repository for literal, high-entropy credentials. I found the Shiro default rememberMe cipher key "kPH+bIxk5D2deZiIxcaaaA==", which is a high-entropy, literal value that can be used to decrypt/forge Shiro rememberMe cookies — a real secret material and therefore flagged.

I ignored other obvious non-sensitive or low-entropy examples:

• "root123", "tomcat/tomcat", "admin/admin", simple passwords and default credentials shown as examples — these are low-entropy/setup defaults and are treated as documentation examples per the rules.
• Email addresses, ownerId GUID, and names/placeholder strings (e.g., DB_PASSWORD, YOUR_API_KEY) are not secrets in this context.