Security audit
Security checks for vulnerabilities and agentic risk
This skill is a heuristic product-checking helper with no hidden persistence or credential access, but users should treat its risk labels as advisory rather than proof.
Install only if you want a lightweight advisory checker for product recommendations. Do not rely on its green/yellow/red labels as authoritative security or authenticity proof; confirm important purchases through official retailers, company records, patents, standards, and independent reviews. Configure the host to fetch URLs only after explicit user request.
The skill promises automatic recommendation interception, URL analysis, scoring, and product credibility detection, but the implementation apparently only provides instructions and static link generation rather than real detection logic. This is dangerous because users may rely on a false sense of protection and trust risk labels that are not backed by actual analysis, especially in a security-themed skill.
The skill instructs the agent to use network-capable actions such as web_fetch and to run a verification script, but it declares no explicit tool scope or allowed-tools boundary. In practice this creates an authorization gap: a host agent may permit broader network access than intended, and users cannot easily tell what external access the skill requires.
The trigger for product verification includes broad natural-language patterns such as any message asking whether a product is legitimate, which can match ordinary conversation and silently invoke verification behavior. Overbroad triggers can cause unintended tool use, surprise network activity, and incorrect insertion of security judgments into unrelated user requests.
The URL-analysis mode can trigger on any URL from broad domains and appears not to require an explicit request to analyze the link. This can lead to unsolicited fetching of third-party content, unnecessary network exposure, and analysis of links the user may have shared for unrelated reasons.
The code heuristically maps product names containing CJK characters to the CN market, and later switches the user-facing output language and resources accordingly. This imposes a locale/language choice automatically rather than offering the user an explicit opt-in or neutral default, which can violate language/locale policy expectations.
No suspicious patterns detected.
