Security audit
Security checks for vulnerabilities and agentic risk
This skill appears to be a local logging utility presented as a rating tool, so users could accidentally persist sensitive input on disk.
Review this carefully before installing. Do not use it with secrets, private prompts, customer data, or proprietary text unless you are comfortable with that content being stored in local log files indefinitely. The package should document retention and deletion, honor its configurable data directory, and narrow its command behavior to real rating tasks before routine use.
The skill is presented as a simple rating utility, but its documentation also advertises persistent local storage, activity logging, search across stored entries, and data export. That expanded behavior increases privacy and data-handling risk because users may invoke it expecting a narrow tool, while it silently accumulates and exposes usage data over time.
The phrase 'for everyday use' and 'Use when you need rating' is overly broad and can cause the skill to be invoked in contexts beyond its intended scope. Broad invocation language increases the chance that the agent routes unrelated tasks to a tool that performs local persistence and logging, which can lead to unnecessary data collection or misuse.
The documentation states that data is stored locally and that each command logs activity with timestamps, but it does not warn users about retention, sensitivity, or how to clear that data. This is dangerous because users may unknowingly create a durable activity trail and exportable dataset, potentially exposing behavioral or content information to other local users or downstream processes.
The tool creates a persistent data directory and records arbitrary user-provided content into local log files without clear disclosure or consent. In an agent skill context, users may pass secrets, prompts, tokens, personal data, or proprietary text that then remains on disk and can later be searched or exported.
The script presents itself as a generic 'rating' utility, but its actual behavior is broad local collection, storage, search, and export of arbitrary user input. This mismatch increases the risk that users will disclose sensitive data under false expectations and makes the tool deceptive from a security/privacy standpoint.
The help text advertises operational commands such as export and status, but the case statement first routes those names into logging branches, preventing the implemented helper functions from being reached. This deceptive control flow can mislead users into believing they are invoking safe informational features while actually causing persistent storage of supplied input.
The export feature writes accumulated historical log content into additional files, increasing data proliferation and the chance of accidental disclosure or later exfiltration. Because the skill description and help do not frame the tool as a data collector/exporter, users are not adequately informed of this privacy impact.
The many pseudo-functional commands all accept arbitrary input and append it to persistent logs, creating a broad and unjustified capture surface for sensitive user data. The breadth of commands makes the tool appear useful for many workflows while primarily acting as a local activity recorder.
User input supplied to commands is silently written to local log files with timestamps and echoed back, without any privacy warning. In an everyday utility or agent setting, users may reasonably enter confidential data, making undisclosed persistence a meaningful privacy and data-handling risk.
No suspicious patterns detected.
