Security audit
Security checks for vulnerabilities and agentic risk
This is a markdown-only content repurposing skill that generates platform-specific drafts and does not install code, access accounts, or publish content automatically.
Install only if you want an agent to transform your supplied content into multiple platform drafts. Avoid giving it confidential material unless you are comfortable with that material being rewritten and translated in the chat, and specify desired platforms and languages when you do not want the defaults.
The trigger list contains broad terms like "rewrite for platforms", "adapt for", "改写", and "转发各平台" that can match many ordinary user requests unrelated to this specific skill. Overbroad activation can cause the agent to invoke this skill unexpectedly, override user intent, and generate unsolicited cross-platform transformations or disclosures of user-provided content.
The rules hard-code output languages by platform and require translation/adaptation even when the user did not request translation. This can lead to unintended data transformation and externalization of user content into additional languages, increasing the chance of privacy, compliance, or accuracy issues when sensitive or nuanced material is republished more broadly than intended.
The trigger phrase '帮我一稿多发' is broad and resembles an ordinary user request rather than a narrowly scoped skill invocation. This can cause accidental activation in unrelated conversations, leading the agent to apply the skill when the user did not explicitly intend to invoke it.
The description states that the Twitter Thread output is in English by default, without indicating user consent or configuration. Forcing a language transformation can violate user expectations, create data handling surprises, or produce unintended outputs in multilingual contexts.
表格与后文说明将 Twitter Thread、LinkedIn 固定为“英文”，且未说明用户可选择中文或其他语言输出。这属于自然语言层面的语言/locale 约束，缺少用户 opt-in 或可选项，可能违反语言选择政策。
The trigger phrases are broad, natural-language requests such as '帮我出一个公众号版本' and '改写成小红书风格', which can plausibly appear in ordinary conversation outside an explicit skill-invocation context. This creates an unintended activation risk where the skill may rewrite or transform user content without clear consent, especially in multi-skill or agentic environments that route based on loose matching.
The skill can reportedly activate from either pasted source text or just a topic, but the boundaries for each mode are not clearly defined. Ambiguous activation criteria increase the chance of unintended execution or misclassification of ordinary content as a skill request.
The markdown states as a fixed platform rule that 'Twitter 是英文', which imposes a language choice in the skill's behavior. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless clearly documented as an optional or justified locale constraint.
No suspicious patterns detected.
