Security audit
Security checks for vulnerabilities and agentic risk
This is a simple Beijing weather advice skill; it may activate too easily on morning greetings but shows no hidden, persistent, or credential-seeking behavior.
Installers should understand that this skill is designed for Chinese-language Beijing morning weather briefings and may run on simple morning greetings. Consider narrowing its trigger if you do not want weather lookups from generic greetings.
The skill metadata says it should be used for very common greetings like “早安/早上好,” which are broad triggers that can cause the skill to activate when the user is merely greeting rather than requesting weather, clothing, or food advice. This can lead to unintended tool use and irrelevant responses, especially in systems that auto-route based on skill descriptions.
The invocation section explicitly prioritizes this skill for generic morning greetings, again without requiring user intent tied to weather or Beijing-specific advice. In an agent environment, this increases accidental invocation risk, which can override a user's actual request and cause unnecessary external lookups.
The instruction to always answer in Chinese imposes a fixed language choice without checking the user's language preference, which may conflict with locale or accessibility expectations. While not a classic security bug, it can cause policy and UX failures if the user spoke another language or the host system expects language continuity.
The example conversations and assistant output are entirely in Chinese, beginning with Chinese-only user greetings and responses. For a general skill example file, this suggests a fixed language expectation without any visible opt-in, alternative locale, or justification that the skill is region-specific.
No suspicious patterns detected.
