Security audit
Security checks for vulnerabilities and agentic risk
This is a single Markdown-only insurance marketing drafting skill with disclosed compliance warnings and no code, persistence, credentials, or automatic publishing behavior.
Before installing, treat this as a drafting aid only. Do not enter real customer personal or health data, verify any regulatory claims and statistics against current official sources, and require qualified compliance/legal review before using generated insurance marketing content publicly.
The trigger keyword list contains broad phrases like '生成文案', '合规审核', and '产品推广' that are common across many benign conversations and domains. In a shared assistant, such ambiguous triggers can cause accidental activation, exposing users to unintended regulated-content workflows or causing the model to treat ordinary requests as insurance marketing tasks.
The skill explicitly states that users can enter requests in any format and that the assistant will infer the stage automatically, which creates overly broad activation semantics. In an agent environment, this increases the chance the skill is invoked on unrelated free-form conversations and begins producing regulated marketing content without deliberate user intent, raising risks of misfire, policy bypass, or unsafe context mixing.
Although the document includes many compliance notes, the main workflow encourages creation of urgency-based and emotionally manipulative insurance marketing copy before the compliance review stage. Because this is a regulated financial-sales context, normalizing '限时紧迫', fear appeals, and '蹭热点' tactics can lead users to generate misleading or coercive promotions that may only be weakly corrected later.
No suspicious patterns detected.
