Security audit
Security checks for vulnerabilities and agentic risk
This skill claims to draft emails and messages, but its instructions include broad triggers and unrelated merged content that could make it activate or respond outside that purpose.
Review before installing. The skill does not show code execution, credential access, persistence, or exfiltration, but its published instructions are poorly scoped and mixed with unrelated material; install only if you are comfortable with a drafting skill that may activate on generic writing requests and may carry unrelated prompt context.
The trigger phrases are extremely broad, including generic requests like '写邮件', '写消息', 'draft', and '写个XX', which are common in normal conversation. Such broad matching can hijack ordinary user requests, causing unintentional skill activation and exposing user content to the wrong prompt context or workflow.
The display name and description are presented only in Chinese, and the remainder of the skill instructions are also written exclusively in Chinese, with no indication that users may choose another language. This can violate language/locale policy when a skill implicitly enforces a language without opt-in or documented regional justification.
The manifest advertises a dedicated drafting assistant, but the body contains a heterogeneous collection of unrelated topics, video references, and auto-merged material. This mismatch undermines trust boundaries and can cause the platform or user to invoke a skill for one purpose while hidden or irrelevant instructions influence behavior for another.
The documented purpose of the skill is quick email/message drafting, but most of the file is filled with unrelated merged 'learning' content and foreign sub-skills. This creates a skill-identity mismatch that can cause the agent to activate or behave outside the user’s expectations, increasing the chance of unintended prompt routing, data misuse, or confusing outputs.
The trigger condition activates whenever the user 'needs related help' for the merged sub-skill, but it does not define what 'related' means. This ambiguity expands the activation surface and can cause the skill to engage on loosely connected prompts, especially given the already unrelated imported content.
This second ambiguous 'related help' trigger has the same problem: it allows activation without clear scope or user intent. In the context of a polluted skill file containing unrelated topics, that ambiguity increases the likelihood of accidental invocation and cross-contamination of assistant behavior.
No suspicious patterns detected.
