Security audit
Security checks for vulnerabilities and agentic risk
This skill does what it claims: it sends a XiaoHongShu note-writing request to a disclosed external generation API and returns the generated content.
Install this only if you are comfortable sending the prompt, audience, tone, persona, note type, and image settings to xiaonian.cc for generation. Avoid entering secrets, confidential business plans, regulated data, or private personal information unless you trust that external service's handling of the data.
The skill invokes an external networked API but does not declare any tool scope, permissions, or allowed-tools boundary. This weakens governance and review controls because the skill can transmit user-provided content off-platform without an explicit capability declaration, increasing the chance of unnoticed data exfiltration or unauthorized network use.
The activation description includes broad trigger terms like '生成小红书笔记/小红书文案/笔记', which can cause the skill to activate for generic writing requests beyond the user's clear intent to use this external service. Because the skill sends prompts to a third-party API, overbroad activation raises the risk of accidental invocation and unintended disclosure of user content.
The skill description and workflow state that content is sent to an external API and that no auth is required, but they do not include a clear user-facing warning or consent step before transmitting user requests. This is dangerous because users may unknowingly provide personal, confidential, or proprietary text that is then disclosed to a third party outside the primary system boundary.
The module description states that the script generates XiaoHongShu notes, and the defaults further assume a Chinese-language audience and persona. This appears to impose a specific language/locale context without an explicit opt-in or documented justification in the file.
This markdown file documents a POST endpoint that sends user-supplied task descriptions and related content fields to a remote service, but it does not warn that the data will be transmitted off-tool to an external API. Under the markdown-specific warning rule, skills that may affect user privacy should disclose that behavior.
The example request fixes the audience and stylistic parameters to a specific Chinese-language context, which can imply a forced locale or language preference if used as a default skill behavior. The file does not indicate that users may choose another language or audience, nor that the locale constraint is intentional and justified.
This code sends the user-supplied task description, audience, tone, character, and image-generation settings to a remote HTTP endpoint. Although the module docstring documents the endpoint, there is no explicit user-facing disclosure at the point of execution, such as a print/log message or confirmation, that user content will be transmitted to an external service.
No suspicious patterns detected.
