Security audit
Security checks for vulnerabilities and agentic risk
This is a legitimate web-search skill, but it can auto-trigger broad searches and records raw search activity without clear retention or opt-out controls.
Install only if you are comfortable with broad automatic web-search activation and possible recording of your search queries and visited-result URLs. Avoid using it for sensitive personal, legal, medical, financial, or proprietary queries unless the runtime environment lets you disable or limit logging and require explicit invocation such as /bo-opensearch.
The skill is configured to auto-trigger on very common conversational terms such as '搜索', '查一下', '查查', and 'search'. In an agent environment, this can cause the skill to activate unintentionally during ordinary dialogue and perform live web fetching without the user explicitly opting in, increasing the chance of unnecessary data exposure, prompt-routing abuse, or unexpected external requests.
The skill advertises triggers such as “搜索”, “查一下”, “查查”, and “search”, which are extremely broad, high-frequency phrases that can appear in many ordinary conversations. This can cause unintended invocation of the skill, leading to unplanned web queries, external content retrieval, and possible disclosure of user intent or sensitive prompts to search/fetch backends.
The README describes search functionality but does not prominently warn users that the skill actively fetches real-time external web content. Without clear disclosure, users may unknowingly cause outbound requests or assume the response is purely model-generated, which weakens informed consent and can create privacy, compliance, or trust issues in sensitive environments.
L121-L137 及全文响应模板均以固定中文格式规定输出，而文档同时接受英文触发词“search”（L003），说明该技能面向多语言输入场景。文件中没有提供语言/locale 选择机制，也未声明仅限中文环境使用，因此存在未经用户同意即强制特定语言输出的政策风险。
The skill specifies logging of raw queries, fetched URLs, scores, dispositions, and timestamps, but does not describe user notice, consent, retention limits, or redaction. Because search queries may contain sensitive personal, medical, legal, or proprietary information, retaining this data without clear disclosure increases privacy and compliance risk.
No suspicious patterns detected.
