Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed Xiaohongshu title-writing helper with no hidden execution, data access, persistence, or privileged behavior.
Install this if you want Chinese Xiaohongshu/RedNote-style title suggestions optimized for engagement. Be aware it intentionally uses platform-specific persuasive wording, emojis, urgency, and compliance filters, so it may be unsuitable for neutral, multilingual, or non-marketing writing unless invoked deliberately.
The skill hard-codes a Chinese-platform optimization style ('native, emotional', emoji requirements, anti-'AI-speak' filtering) without checking the user’s language, platform, or consent. This can override user preferences and produce manipulative or non-compliant outputs in the wrong context, especially if the skill is auto-invoked due to the broad trigger.
The trigger condition accepts essentially any raw text or topic, making the skill likely to activate on ordinary user requests that were not intended for title-generation. Overbroad activation can cause prompt hijacking of unrelated conversations, unwanted style transformation, and accidental application of the skill’s persuasive-output behavior outside its intended scope.
The file presents all example categories, inputs, and preferred outputs exclusively in Chinese, which implies the skill behavior is oriented toward a single language. There is no indication that users can opt into this locale or that the skill is intentionally limited to a Chinese-language context.
The docstring states the validator filters titles based on 'Xiaohongshu algorithm preferences', and the hard-coded banned terms and engagement rules are tailored to a Chinese-language content environment. This imposes a locale/language-specific policy in the skill logic without indicating user choice or explaining that the skill is intended only for that region/platform context.
No suspicious patterns detected.
