Security audit
Security checks for vulnerabilities and agentic risk
The skill mostly does what it claims, but it tells the agent to delete prior HR report files with broad wildcard shell commands without clear confirmation or containment.
Review before installing. Use only in an approved local HR workspace, avoid shared or synced folders unless intended, and do not allow automatic regeneration to run the rm -f cleanup commands without confirming the exact directory and files. Prefer versioned report outputs over deletion. Also verify that the scoring keywords and weights match the actual role before relying on rankings.
The command rm -f /目标目录/简历筛选报告*.xlsx is a destructive filesystem operation using a wildcard and a variable target directory. If the directory is misresolved, attacker-influenced, or broader than intended, the agent could delete unrelated files without confirmation.
# 删除该目录下所有之前生成的筛选报告文件
rm -f /目标目录/简历筛选报告*.xlsx
rm -f /目标目录/面试邀请名单*.xlsx
rm -f /目标目录/候选人详细评分依据*.xlsx
The command rm -f /目标目录/面试邀请名单*.xlsx performs wildcard deletion in a user-referenced directory, creating risk of unintended data loss. In an automated agent context, this is especially dangerous because there may be no interactive safeguard to catch an over-broad path or naming collision.
# 删除该目录下所有之前生成的筛选报告文件
rm -f /目标目录/简历筛选报告*.xlsx
rm -f /目标目录/面试邀请名单*.xlsx
rm -f /目标目录/候选人详细评分依据*.xlsx
The command rm -f /目标目录/候选人详细评分依据*.xlsx can erase detailed candidate assessment files based on a broad filename pattern. Because these files contain sensitive HR data, accidental deletion can cause both operational disruption and potential compliance issues around record handling.
rm -f /目标目录/简历筛选报告*.xlsx rm -f /目标目录/面试邀请名单*.xlsx rm -f /目标目录/候选人详细评分依据*.xlsx
### 步骤2：使用验证过的评分数据
The skill processes highly sensitive resume data and writes output files, while elsewhere also recommending deletion of prior files, but it provides no user-facing privacy notice, retention policy, or warning about destructive operations. In practice this can lead to unintended exposure of personal data or accidental loss of prior screening artifacts in shared or synced directories.
The file explicitly centers on Chinese resume parsing and includes Chinese-specific implementation requirements, such as mandatory use of tools for Chinese PDF extraction and Chinese naming assumptions. There is no indication that users can opt for another language or locale, which constitutes a language/locale policy issue under the specified rule.
Line L510 states the skill is applicable to any industry or function and does not preset specific industry vocabulary. However, the documented extraction and scoring logic hardcodes numerous semiconductor, hardware, AI, and market-role keywords, which directly contradicts that claim and biases behavior toward particular domains.
The skill embeds shell-based deletion commands that are not essential to the core resume-screening logic and instruct deletion using a broad user-supplied target directory with wildcards. In an agent setting, this creates a real risk of destructive file operations if the path is incorrect, over-broad, or influenced by untrusted input.
No suspicious patterns detected.