Security audit
Security checks for vulnerabilities and agentic risk
This skill is a simple legal triage helper with broad and Chinese-focused routing, but it does not show hidden execution, data exfiltration, persistence, or privileged behavior.
Use this as general legal triage only. Confirm that it was intentionally invoked for a legal issue, be aware that it appears optimized for Chinese-language inputs, and do not treat its output as a substitute for advice from a qualified attorney in the relevant jurisdiction.
The skill description is broad enough to capture many ordinary help-seeking requests, which can cause the agent to route non-legal conversations into this skill unnecessarily. In a legal domain, over-triggering is risky because users may receive legal-flavored guidance in situations that are not actually legal matters, increasing confusion and the chance of inappropriate reliance.
The example phrases are highly ambiguous and include generic prompts like asking what to do or what one's rights are, without clear legal qualifiers or counterexamples. This makes accidental invocation more likely and can funnel general advice requests into a legal-consultation workflow, which is especially sensitive because users may interpret the response as authoritative despite the disclaimer.
The description is broad enough to match a wide range of ordinary legal-help queries, which can cause the skill to be invoked more often than intended. In a legal domain, overbroad routing is risky because users may receive generalized or inappropriate guidance for regulated, high-stakes matters such as criminal, immigration, family, or financial disputes.
The trigger list includes very generic phrases such as '怎么办', '我的权利', and '合法吗', which can appear in many ordinary conversations unrelated to requesting this specific skill. In a legal-consultation skill, broad activation is risky because it can silently route users into legal-guidance behavior in ambiguous contexts, increasing the chance of inappropriate advice, privacy exposure, or unwanted skill invocation.
The classifier only checks for Chinese keywords, so users asking legal questions in other languages will be silently misclassified as 'general' and may not receive the appropriate specialist suggestions. In a legal-consultation skill, this can degrade guidance quality and route users away from relevant help, which is safety-relevant because legal triage depends on accurate issue identification.
The common scenarios and usage examples are written exclusively in Chinese while the rest of the document is in English, which implicitly steers the skill toward a specific language context. There is no statement that the user may interact in any language or that Chinese-only examples are intentional for a region-specific deployment.
All listed trigger phrases are in Chinese, and the manifest does not indicate that users may invoke the skill in other languages or opt into a Chinese-only experience. This creates a natural-language locale constraint without user choice or a documented region-specific justification.
This JavaScript test file contains natural-language examples only in Chinese for core legal-consultation inputs. That can indicate an implicit language restriction without user opt-in or explicit documentation justifying the locale constraint.
No suspicious patterns detected.
