Security audit
Security checks for vulnerabilities and agentic risk
This skill locally turns dependency-upgrade notes into a structured briefing and does not show hidden network, persistence, or destructive behavior.
Install only if you want Chinese-language dependency-upgrade briefings. Provide dependency names, version changes, and changelog summaries rather than broad sensitive directories; if using --output, choose an intended local output path.
声明描述的是一个面向“依赖升级收益/风险/回滚/业务影响”的专项分析能力；但代码本身并没有实现任何依赖解析、版本比较、升级影响评估、回滚规划或兼容性判断逻辑。相反，它是一个由 spec 驱动的通用报告框架，可扫描目录、分析 CSV、匹配安全风险模式、检查 Skill 文件完整性与 frontmatter。虽然 structured_brief 模式可以把任意输入组织成报告，表面上可能被用于撰写依赖升级说明，但这只是泛化文本模板能力，不足以证明其主要用途与声明一致。其实际能力范围明显更广且与“依赖升级分析”并不特定对应，因此属于描述与行为的实质性不匹配。
The code performs regex-based scanning for secrets, internal/private URLs, and risky shell patterns across arbitrary files or directories, which is unrelated to dependency-upgrade analysis. In this skill context, that makes the capability more dangerous because it enables broad local content inspection and extraction of sensitive snippets from user files under a misleading skill description.
The dispatcher enables multiple operational modes such as directory auditing, CSV auditing, pattern scanning, and skill auditing that materially exceed the declared purpose of explaining dependency upgrade benefits, risks, rollback, and business impact. This scope expansion creates an unnecessary capability surface for inspecting arbitrary local files and repositories, which is dangerous in an agent-skill context because users may invoke it with sensitive paths under the assumption it is narrowly focused.
The trigger examples are broad, conversational phrases such as asking whether a dependency upgrade is worth doing or requesting a business-impact explanation. In skill-routing systems, overly generic triggers can cause accidental invocation outside the intended dependency-upgrade context, which may lead to irrelevant outputs, confusion, or misuse of the skill in adjacent workflows.
The entire self-check document is written in Chinese, including the title and all audit content, with no indication that other languages are supported or that Chinese is a justified locale constraint. Under the policy rule, a skill that forces a specific language without user opt-in can be a natural-language policy violation.
The skill declares no explicit tool scope or permissions, yet the content instructs use of python3 and references reading templates/specs and producing output files, implying shell, file-read, and file-write capabilities. Without an allowlist, an agent may grant broader-than-necessary execution and filesystem access, increasing the chance of unintended command execution or data exposure if the skill is invoked in a sensitive workspace.
The description forces a mixed Chinese/English presentation for routing and usage guidance, but does not state that the user may choose their preferred language. Under the policy, language constraints should be optional or explicitly justified; here no opt-in or locale justification is provided.
The template content and headings are entirely in Chinese, and the title indicates a fixed Chinese output template. This can constitute a language-policy issue because the skill appears to require a specific language without offering the user a language or locale choice.
The generated report strings in make_structured_report are fixed in Chinese, which imposes a specific language on users. The file does not provide any option, configuration, or documented opt-in for locale selection, so this is a natural-language policy concern applicable to code files.
directory_report, csv_report, pattern_report, parse_frontmatter, and skill_audit all emit Chinese-language headings or messages directly. Because the script consistently forces one language in user-visible output and error text without offering alternatives, it violates the language/locale policy criterion.
The skill_audit workflow audits package structure and parses SKILL.md metadata, which is unrelated to dependency-upgrade briefing and expands the tool into a generic repository/compliance inspector. While less severe than secret scanning, it still broadens access to local project contents and violates least functionality for the declared skill purpose.
The CLI description presents the tool as a local support script, but the implementation includes substantive audit and scanning behaviors over directories, CSV files, and skill packages. This misleading presentation increases the chance of unsafe operator trust and accidental use on sensitive data because the real capabilities are broader than the label suggests.
The entire example output is written in Chinese and the title indicates a Chinese-language briefing format, but there is no natural-language indication that the user can choose another language or locale. This can violate a language/locale policy if the organization expects user-facing output language to be optional unless explicitly justified.
