Security audit
Security checks for vulnerabilities and agentic risk
This is a simple personal-assistant instruction skill with broad but disclosed task, communication, scheduling, and note-taking guidance, and no executable code or hidden install behavior.
Reasonable to install if you want broad personal-assistant behavior. Before using it with private messages, calendars, documents, or business information, set clear expectations for what it may store, when it must ask before acting, and what sensitive content should stay out of notes or memory.
The skill is defined as a generic 'Assistant' with a broad description and no activation boundaries, exclusions, or negative examples. That makes accidental or over-broad invocation more likely, which can route arbitrary user requests into a skill that is authorized to manage tasks, communications, and scheduling, increasing the chance of unintended access to sensitive workflows or data.
The instructions encourage summarizing threads, drafting replies, keeping notes on preferences, remembering prior context, and managing information, but they do not include guardrails for sensitive data handling, least-privilege use, or restrictions on confidential content. In an assistant context, this increases the risk of exposing, retaining, or acting on personal, business, or regulated information without adequate user awareness or approval.
No suspicious patterns detected.
