Security audit
Security checks for vulnerabilities and agentic risk
This skill is coherent legal-analysis tooling, but it automatically persists potentially confidential case reports to the workspace without a clear opt-in save step.
Install only if you are comfortable with the agent processing Chinese legal matter materials and saving a report in the workspace. Use the explicit /lawyer-six-steps command, avoid pasting unnecessary confidential material, and confirm where any report will be written before proceeding.
The natural-language triggers include broad phrases such as '帮我分析一个案件' and '案件预判', which can match ordinary conversation rather than an explicit request to invoke this skill. In an agent environment, this can cause unintended activation and make the model process sensitive legal-case information under the wrong workflow, increasing privacy and safety risk.
The additional trigger examples remain generic and semantically broad, especially phrases like '帮我分析一个案件', '案件预判', and '办案分析'. Because this is a legal-analysis skill handling potentially confidential matters, accidental triggering is more dangerous than in a low-stakes domain: it may route sensitive user content into a specialized workflow without deliberate consent.
The skill description and instructions are written to operate specifically in Chinese legal practice context, but they do not explicitly offer the user a language choice or state that the locale restriction is mandatory for a region-specific compliance reason. Under the policy, forcing a specific language without opt-in can be a natural-language policy violation.
The natural-language triggers are broad terms such as '案件分析' and '要件审查', which can appear in ordinary legal conversation and may cause the skill to activate unintentionally. Because this skill performs structured legal analysis and ultimately instructs file output, accidental activation can trigger unwanted processing of sensitive matter descriptions and downstream side effects.
The skill directs the agent to write a complete case analysis report into the workspace by default, which is a side effect beyond normal chat-only analysis. In a legal-analysis context, that report is likely to contain sensitive client facts, claims, evidence summaries, and risk assessments; automatic file creation can therefore leak or persist confidential data in an unintended location.
The skill instructs writing a report file but the markdown description does not clearly warn users that local workspace files may be created. This reduces informed consent and is especially risky for a lawyer-facing skill, where generated reports may include privileged or highly sensitive client information that users may not expect to be persisted.
The prompt instructs use of a Chinese litigation analysis workflow for the case, which imposes a specific locale/jurisdiction context by default. The file does not indicate that users can choose a different language or legal jurisdiction, nor that this region-specific constraint is presented as an explicit opt-in.
All instructions and required outputs are specified in Chinese, which effectively forces a specific language for use of the skill. The file does not state that the skill is region-specific or provide any opt-in or alternative language option, so it conflicts with the language-choice policy.
The skill explicitly instructs the agent to enter a final report output stage and only tell the user '已写入并打开', which implies writing/opening a file while suppressing meaningful user-facing disclosure. That creates an unsafe side effect boundary: the agent may modify local artifacts or trigger UI/file actions without explicit contemporaneous consent or visibility.
This markdown file presents all headings and template instructions exclusively in Chinese, which imposes a specific language on users. Under the policy rule, forcing a language without user opt-in or a documented regional justification is a natural-language policy violation.
No suspicious patterns detected.
