Security audit
Security checks for vulnerabilities and agentic risk
The skill is mostly coherent device-control documentation, but it enables immediate hardware shutdown from natural-language intent without a confirmation step.
Review this before installing if the controlled device is important or shared. The main issue is not hidden malware, but that a user phrase like 'turn off the device' can become an immediate shutdown command without an explicit confirmation step in the skill instructions.
The protocol defines an immediate power-off command with no confirmation, safety interlock, or user warning. Because shutdown is disruptive and potentially irreversible in the moment, an ambiguous or maliciously induced invocation could cause denial of service, interrupt active sessions, or terminate critical device functions without recourse.
The skill explicitly supports a power_off action but does not warn that the command can immediately interrupt service or terminate an active session. In a device-control skill, omission of a confirmation or user-facing warning increases the chance of accidental shutdowns and abuse through ambiguous or misinterpreted natural-language requests.
The skill says it should be used whenever natural language implies device control and even allows custom command expansion, but it does not define strict trigger boundaries, authorization checks, or disambiguation rules. In a device-control context, broad activation increases the risk of unintended command generation from ambiguous phrasing, indirect prompt content, or mixed-topic conversations.
The protocol exposes commands for emotion display and device vibration that are outside the stated manifest scope of common hardware controls. This scope drift increases the chance that an agent or integrator will invoke undocumented or unexpected behaviors, weakening least-privilege expectations and potentially enabling user-manipulative or unsafe physical actions.
The natural-language description and invocation examples are entirely in Chinese, which can imply a locale-specific interaction policy. Because the file does not state that the skill is intentionally limited to Chinese users or offer language flexibility, this may violate the language/locale policy criterion.
No suspicious patterns detected.
