Security audit
Security checks for vulnerabilities and agentic risk
This skill is a markdown-only search planning helper that routes research sources and does not execute searches, install code, persist changes, or handle credentials.
Installing this skill is reasonable if you want a Chinese-language source-routing planner. Review its local_file and cache recommendations before using them in sensitive workspaces, and consider fixing the counter-evidence schema mismatch so evaluation and investigation tasks trigger the intended adversarial search planning.
The trigger condition is broad enough that ordinary research tasks could invoke this skill even when source routing is unnecessary. Over-triggering a routing-only skill can distort workflow selection, waste search budget, and suppress more appropriate task-specific skills, though it does not directly create code-execution or data-exfiltration risk.
L003 的描述全文使用中文并将该技能定义为“LLM通用来源路由技能”，但文件中没有说明这是仅面向中文用户或特定区域的技能，也未给出可切换语言的选项。对于“通用”技能，默认强制单一语言会构成语言/locale 选择上的自然语言策略问题。
The counter-evidence trigger logic is internally inconsistent: it checks task_type = evaluate or investigate, but those values belong to goal_mode per the declared input schema. This schema/rule mismatch can cause the router to skip or incorrectly enable adversarial verification, weakening decision quality and increasing confirmation-bias risk in downstream searches.
This markdown file is entirely written in Chinese, including the title, instructions, and examples, with no indication that language selection is optional or that the skill is region-specific. The policy for natural-language content requires flagging language or locale constraints when a specific language is effectively enforced without user opt-in.
The file content is written entirely in Chinese and the title is bilingual, but the operational instructions themselves are only provided in Chinese. This can impose a language/locale constraint on users or downstream agents without any explicit opt-in or justification that the skill is region-specific.
No suspicious patterns detected.
