Security audit
Security checks for vulnerabilities and agentic risk
This skill is a mostly coherent Xiaohongshu publishing helper, but it can publish to a real account without an explicit final confirmation and includes loosely scoped account/profile actions.
Review before installing. Use this skill only when you intentionally want an agent to operate an already logged-in Xiaohongshu creator session, and require a manual confirmation before any publish click. Avoid using the broad account-management or profile-fetching parts unless you have a clear, authorized reason.
The trigger description includes broad language such as managing a Xiaohongshu account, which can cause the skill to activate for requests beyond publishing. In an automation context, over-broad invocation increases the chance of the agent taking sensitive account actions the user did not specifically intend.
The workflow proceeds to clicking the live publish button without requiring an explicit confirmation step immediately before posting. Because publishing is an external side-effect on a real user account, omission of a final consent gate can lead to accidental or premature public posting, reputational harm, or policy-violating content being published.
The skill is scoped as a publisher for posting notes, but it also documents fetching profile information from Xiaohongshu user pages. That expands the capability into profile data collection unrelated to the core task, which can enable unnecessary scraping of account metadata and creates privacy and misuse risk beyond expected publishing behavior.
The profile-access instructions tell the agent to open a user profile page and extract follower counts and note counts, but they do not include any privacy warning, scope limitation, or consent guidance. Even if the data is publicly viewable, automating its retrieval can normalize collection of account information outside the skill's main publishing purpose.
No suspicious patterns detected.
