Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed retail sales reporting helper, but users should configure data access, recipients, and schedules carefully.
Before installing, confirm that report_config points only to intended POS/ERP data, uses an approved recipient list, and has schedules enabled only for reports you actually want sent. Treat generated reports as sensitive business data, especially revenue, inventory, anomalies, and staff-performance sections.
The on-demand trigger is defined as "Any data question," which is overly broad for a skill that can access POS/ERP data and generate reports. This can cause unintended activation on casual conversation, leading to unnecessary data retrieval or exposure of business metrics without clear user intent.
The delivery section instructs the skill to send reports through external channels such as WeCom or Telegram, but it does not warn the user that operational business data will be transmitted outside the immediate chat context. Because the reports may include revenue, inventory, anomalies, and staff performance data, omission of a disclosure or consent mechanism increases the risk of accidental data sharing to inappropriate destinations.
The skill states that it also triggers for scheduled automatic report delivery, but it does not define guardrails such as enabled schedules, authorized recipients, required configuration state, or failure behavior. In a reporting skill that handles commercial data, unclear activation boundaries can result in unintended or repeated delivery of sensitive business information.
The metric labels and display strings prescribe Chinese-language output such as "昨日对比", "上周对比", and other Chinese labels, and the display formatting section further mandates a Chinese sentence template. Because the document does not offer an alternative language or user opt-in, this appears to impose a specific language/locale choice.
No suspicious patterns detected.
