Security audit
Security checks for vulnerabilities and agentic risk
This skill is a coherent health-planning helper that uses sensitive health inputs for its stated purpose, with no evidence of hidden execution, exfiltration, persistence, or destructive behavior.
Before installing, understand that this skill may ask for sensitive health information such as conditions, medications, supplements, labs, body metrics, and routines. Share only what is necessary, keep generated reports/reminders in locations you trust, and treat the output as lifestyle coaching that may require clinician review for medical decisions.
声明描述的是一个多模块、长期健康管理与长寿规划技能，而代码片段的实际功能非常聚焦，只覆盖声明中的一个子能力：每日饮食/营养均衡分析与热量缺口跟踪。它不会处理病史、生活习惯全量建模、保健品审查、自动周期报告或提醒调度等能力。因此描述显著宽于实际代码，构成能力层面的不匹配。虽然代码行为与声明中的“每日饮食营养分析（含热量缺口）”部分一致，也未发现额外越权或隐藏能力，但整体 declared purpose 并不能准确代表该代码块的实际范围。
声明描述的是一个较完整的健康长寿规划与行为改变工具，核心能力包括行动计划生成、风险评分、自动周/月报、保健品安全审查、饮食营养分析、热量缺口追踪和提醒。实际代码的主要功能明显更窄：它只读取输入JSON，基于固定规则计算风险评分，并生成周报/月报Markdown。虽然“风险评分”和“周报/月报”与声明部分一致，但多个重要宣称能力未在代码中出现。尤其是饮食营养分析、热量缺口、提醒机制、交互式信息收集、以及真正的supplement安全审查都没有实现。代码中的保健品部分只是统计已有status标签数量，而不是根据成分或用药背景进行安全判断。因此描述对能力范围有明显夸大，属于描述与实际行为不符。
声明描述的是一个综合性健康/长寿规划与分析技能，包含个体信息采集、计划生成、风险评估、营养分析、补剂安全审查和周期报告等多项高级功能。实际代码只处理提醒排程：解析时间、扩展间隔规则、按profile映射星期、过滤quiet hours，并输出每周提醒清单与统计。虽然提醒类型与声明中的部分提醒场景（饮水、站立、睡眠、运动、保健品）有一定相关性，但代码并未实现大部分核心承诺能力，因此描述与实际行为存在明显不匹配。
The skill instructs collection of sensitive health data including diseases, medications, supplements, body metrics, and potentially pregnancy-related information, but provides no explicit privacy notice, retention limits, or user consent step. In a health context, this is dangerous because users may disclose regulated or highly sensitive information without understanding how it is processed, stored, or shared.
The skill references local files and explicitly instructs running scripts that read JSON input and write Markdown output, but it declares no tool scope or permissions. This creates an authorization and transparency gap: an agent may be induced to access files or produce artifacts without the skill manifest clearly constraining or disclosing those capabilities.
The invocation wording is broad enough to capture many ordinary health requests, increasing the chance the skill activates in contexts involving sensitive medical, nutrition, or medication discussions. Overbroad matching can route users into a workflow that collects high-sensitivity data and may generate quasi-medical guidance without sufficient gating or consent.
The skill proposes recurring reminders, auto reports, and potentially platform-ready schedules that may create persistent records or trigger actions on external systems, yet it lacks explicit warning or consent language. This can lead to unintended disclosure of health-related routines or generation of health-tracking artifacts on third-party platforms.
This JSON contains user-facing report strings in Chinese across multiple fields, while no accompanying indication in the file shows that the user selected Chinese or that the report is intentionally region-specific. Under the policy, forcing a specific language without opt-in is a natural-language policy concern.
The intake template asks for highly sensitive health, medical, and lifestyle information, including diagnosed conditions, medications, lab values, and family history, but provides no privacy notice, purpose limitation, retention guidance, or user consent language. In a skill designed to collect longitudinal wellness data and reminders, this omission increases the risk of over-collection, inappropriate storage or sharing, and user harm if the data is exposed or mishandled.
The file includes multiple user-facing reminder messages written in Chinese, while the surrounding document is otherwise in English and does not state that the skill is China-specific or requires Chinese output. This creates a language policy issue because the skill appears to impose a specific language on users without opt-in or justification.
This JSON file contains user-facing reminder messages in Chinese while providing no indication that the skill is China-specific or that the user can opt into this locale. Under the policy, forcing a specific language without user choice or documented regional justification is a natural-language policy violation.
The example sets the timezone to Asia/Shanghai, which implies a locale-specific default, and the reminder messages are written in Chinese (喝水 300ml, 起身活动 3-5 分钟, 进入睡前流程) without indicating that language or locale is optional. This can violate language/locale policy when a skill implicitly assumes a specific user language or region instead of offering choice or documenting the constraint.
This code file contains multiple Chinese-only status strings, suggestions, headings, and the final disclaimer, which forces a specific language for end users. Under the policy, language constraints should either be optional, user-selectable, or clearly justified as region-specific.
This Python file embeds Chinese-only labels, headings, and risk messages throughout the generated report, such as contributor names, risk bands, and markdown section titles. The skill does not offer any language selection or document a justified region-specific constraint, which matches the language/locale policy violation criteria.
The script defines Chinese-only weekday and reminder type labels and later uses Chinese headings and content when generating markdown. This forces a specific language/locale in the user-facing output without opt-in, which matches the language-policy violation criteria for all file types.
The top-level description is primarily in Chinese but includes English instruction text, and the rest of the document is written in English. This may create a language/locale mismatch for users or downstream systems without explicitly asking which language the user prefers.
This manifest presents the skill name and description in Chinese, but the default prompt is fixed in English. That creates a language-policy concern because the skill appears to impose a language choice without explicit user opt-in or any documented locale justification.
This JSON sample mixes English field names with Chinese meal-name values such as the entries on L23, L33, L43, and L53. Because SQP-3 applies to all file types and flags language/locale policy issues, hardcoding a specific language in user-facing content without opt-in or documented regional scope is a natural-language policy concern.
No suspicious patterns detected.
