Security audit
Security checks for vulnerabilities and agentic risk
This is a plain Chinese writing-assistant skill with no code execution, credential use, network access, persistence, or hidden high-impact behavior.
Installers should understand this skill is aimed at Chinese writing workflows and may activate on common Chinese requests like writing titles, copy, or emails. It does not appear to access private data or run commands.
The manifest description, title, and all listed trigger phrases/examples are exclusively in Chinese, which implies the skill is intended to operate only in Chinese. The file does not provide any user opt-in, alternative language option, or justification for a locale-specific restriction, which matches the language/locale policy violation criteria.
The entire skill, including title, triggers, prompt template, and example dialogue, is written only in Chinese and implicitly expects Chinese-language use. There is no opt-in language choice or stated justification that this is a Chinese-only or region-specific skill.
The trigger phrases 写文案 and 帮我写文案 are broad, natural-language expressions that can easily appear in ordinary conversation, making unintended skill activation plausible. While this is not an exploit in itself, it can cause prompt routing mistakes, unexpected behavior, or invocation of this skill when the user intended a general request.
The skill is authored entirely in Chinese and its examples/output format assume Chinese output, without stating a locale limitation or allowing the user's language preference to control responses. This can cause incorrect or non-compliant behavior for users expecting another language, which is a robustness and policy-handling issue rather than a direct code-execution risk.
The trigger phrases '写邮件' and '帮我写邮件' are very broad and likely to match ordinary user requests in many contexts, causing this skill to activate unintentionally. Over-broad routing can misdirect user input into a fixed template, leading to incorrect handling of requests and making downstream prompt behavior easier to influence or abuse at the orchestration layer.
文件整体以中文定义标题生成器的名称、触发词和提示模板，默认输出也为中文，但未说明这是仅面向中文场景的专用技能，也未给用户提供语言/locale 选择。根据规则，若技能强制特定语言而无用户 opt-in，可构成自然语言策略问题。
The trigger phrases are short, generic everyday expressions that can easily appear in normal conversation, increasing the chance the skill activates when the user did not explicitly intend to invoke it. In an agent environment, unintended activation can cause context switching, unexpected prompt shaping, or disclosure of user content to the skill logic.
No suspicious patterns detected.
