Security audit
Security checks for vulnerabilities and agentic risk
This is a small static code-review helper with disclosed security-linting behavior and no executable payload, persistence, or hidden privilege request.
Installers should know this skill may consult external security references while reviewing code. It appears safe as a lightweight static-review aid, but users handling private code should ensure the agent does not send sensitive snippets, secrets, or proprietary details to external pages or searches.
The README describes the skill entirely in Chinese and presents its behavior/output conventions in Chinese without indicating that users may choose another language. This is a natural-language locale policy issue because it imposes a specific language by default with no opt-in or documented region-specific justification.
The manifest description and all user-facing instructions are written only in Chinese, and the output format specifies Chinese risk labels without any opt-in or alternative language option. This creates a language policy concern because the skill appears to enforce a specific locale regardless of user preference.
The trigger condition '当用户要求审查代码时' is overly broad and can activate for ordinary code-review requests, not just security linting. Overbroad activation can cause the skill to run in unintended contexts, potentially applying network-enabled behavior or security-focused handling where it was not requested, increasing attack surface and the chance of unsafe tool use.
The skill is a local code-review/linting assistant, but it explicitly instructs the agent to use browser or web_fetch to consult external security guides. This expands the skill’s capability surface to network access without a clear necessity, creating risks such as unintended data exfiltration, prompt-injection exposure from fetched pages, or privacy leakage if code context is sent outward.
No suspicious patterns detected.
