Security audit
Security checks for vulnerabilities and agentic risk
This skill opens a requested website in the default browser and does not show hidden data access, persistence, or destructive behavior.
Install only if you want a simple Chinese-language URL opener. Treat it like a browser shortcut: only let it open links you asked for or recognize, and do not rely on it to choose Safari or Chrome specifically.
代码的核心功能与“打开指定网址”基本一致，因此不存在完全不同的主用途，也没有发现额外敏感或未声明的资源访问行为。但描述中明确写到“支持 Safari/Chrome 浏览器”，容易让人理解为可选择或控制这两个浏览器之一；实际实现仅使用 webbrowser.open(target_url)，通常是调用系统默认浏览器，并没有 Safari/Chrome 的显式适配、选择或控制逻辑。此外，描述称其为“macOS 浏览器自动化工具”，而代码并无明显的 macOS 专属处理，也没有除打开 URL 之外的自动化操作。因此描述与实际行为存在一定程度的不准确，属于描述能力强于实际实现的 mismatch。
The skill name, description, tags, and usage documentation are entirely in Chinese, with no indication that users may choose another language or that the skill is intentionally limited to a Chinese-speaking audience. Under the stated policy, forcing a specific language without opt-in or clear justification is a natural-language policy concern.
The natural-language text in the docstring and returned messages is fixed to Chinese, which can force a specific language experience on users without opt-in. The policy requires avoiding language or locale constraints unless the skill offers a choice or clearly documents a justified regional limitation.
The user-facing name, descriptions, example input, and success message are all presented only in Chinese, with no indication that the skill supports other languages or that Chinese-only behavior is an intentional, justified regional limitation. This is a natural-language policy concern because the manifest imposes a locale/language choice without opt-in.
The manifest describes a broadly callable browser-opening capability that can open arbitrary URLs, but it does not define clear trigger constraints, allowed domains, or user-confirmation boundaries. In an agent environment, this can enable prompt-induced browsing to attacker-controlled sites, phishing pages, or sensitive local/network resources, increasing the chance of unintended external actions.
The skill opens whatever URL is provided immediately, with no confirmation, allowlist, or safety check before triggering the browser action. In an agent setting, this can cause unwanted navigation to phishing, tracking, or other untrusted sites if the input is manipulated or misunderstood, even though the code itself appears intended for convenience rather than abuse.
No suspicious patterns detected.
