Security audit
Security checks for vulnerabilities and agentic risk
This is a local trade-document consistency checker with no evidence of network access, credential use, persistence, or destructive behavior.
Install only if you are comfortable with a local, rule-based helper for Chinese or bilingual trade-document text. Do not rely on a clean result as bank or legal assurance; the tool only compares fields it can extract from your supplied documents and explicitly omits deeper arithmetic, date, LC-limit, and document-set checks.
主要目的基本一致：都是做外贸单证“单单一致”检查，并且代码没有发现越权、联网、付费、注册等未声明行为。但描述与实现存在明显能力层面的落差。描述把免费版说成覆盖五大类检查，其中金额/币种、数量/件数/重量、港口/船名/唛头看起来像明确交付的专项能力；而代码实际核心只有‘基于标签抽取后的通用跨单据字段差异检测’和‘模板占位符残留’。虽然通用比较在一定条件下可能覆盖金额、数量、港口、船名等字段，但实现依赖非常朴素的“标签:值”模式提取，既没有更深层规则，也没有分别实现这些类别的专项校验逻辑，因此描述比真实能力更强。故应判定为存在描述与行为不完全匹配。
Referenced artifact was not completely inspected
node scripts/run.mjs --sample
Referenced artifact was not completely inspected
node scripts/run.mjs --sample
Referenced artifact was not completely inspected
node scripts/run.mjs --sample
The file advertises that the free version performs five classes of checks, including amount/currency, quantity/packages/weight, and port/vessel/marks consistency, but the implementation only runs cross-document field extraction plus placeholder detection. This creates a security-relevant integrity gap: users may rely on claimed screening coverage for trade-document discrepancies that are not actually being validated, leading to false assurance and missed errors in high-stakes banking/shipping workflows.
The inline documentation states the free tier only performs two check types, while the returned scope still advertises a broader set. Conflicting statements about validation coverage can mislead downstream systems or users into believing more controls were applied than actually ran, which is dangerous for a compliance-oriented document-checking skill.
The file's user-facing description, usage text, and runtime messages are entirely in Chinese, which imposes a specific language on users. There is no indication that users may choose another language, nor is there a documented region-specific justification for the locale constraint.
The field extraction logic is documented and implemented with a Chinese-centric label set, and the file-level comments and sample input are also entirely Chinese-focused. This can amount to a locale/language constraint without explicit user opt-in or documentation that the skill is region-specific only.
The JSON value contains bilingual document text with Chinese labels and English headings, but there is no indication that language selection is optional or user-controlled. Under the stated policy, forcing or assuming a language/locale without opt-in can be a natural-language policy issue.