Security audit
Security checks for vulnerabilities and agentic risk
This coaching skill has a coherent purpose, but it silently processes sensitive chat-history exports and sets persistent review reminders without clear consent or controls.
Review this carefully before installing. Only provide chat exports after removing secrets, private third-party information, and anything you do not want summarized into memory. The skill should ideally ask before ingestion, show what it extracted, and let you cancel or delete stored review cards and scheduled reminders.
The entire skill definition mandates behavior and sample outputs in Chinese and includes stylistic constraints without indicating that the user can choose another language. This can violate language/locale policy because it forces a specific language experience without opt-in or documented justification.
The skill tells the agent to receive exported historical chat records, process them 'in the background,' and respond only with a terse success message, without warning users about the scope of ingestion, retention, or sensitivity of the data being analyzed. Historical chats often contain secrets, personal data, or third-party information, so silent background processing increases privacy risk and reduces informed consent.
The skill tells the agent to receive exported historical chat records, process them 'in the background,' and respond only with a terse success message, without warning users about the scope of ingestion, retention, or sensitivity of the data being analyzed. Historical chats often contain secrets, personal data, or third-party information, so silent background processing increases privacy risk and reduces informed consent.
No suspicious patterns detected.
