Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed execution-progress logger with no network, credential, persistence, or destructive behavior, but users should note its broad triggers and verbose process output.
Install only if you want Chinese step-by-step progress messages. Consider narrowing or disabling automatic triggers such as '过程' if accidental activation would be annoying, and avoid using it to reveal hidden chain-of-thought, private prompts, credentials, or sensitive intermediate data.
The trigger phrase '过程' is extremely broad and is likely to match many ordinary prompts, making accidental activation highly probable. In this skill, accidental activation is especially risky because the stated behavior includes exposing internal execution steps and reasoning-related content to the user.
The skill documentation is written entirely in Chinese and presents the interaction/output style as the intended behavior, with no indication that users may choose another language. This can violate a language/locale policy when a skill imposes a specific language without explicit opt-in or justification.
This section directs the agent to reveal internal execution details step by step, including skill selection, model invocation, waiting states, returned results, and reasoning. Such transparency can leak implementation details, facilitate prompt probing, and expose data or decision pathways that should remain abstracted from end users.
The skill explicitly instructs the agent to provide '推理过程' and to rule out alternatives without requiring explicit user opt-in. Revealing reasoning/process details can expose sensitive internal deliberation, policy logic, or hidden system behavior beyond what should normally be user-visible.
The auto-trigger behavior is keyed off common conversational phrases like '分步', '一步一步', and '执行过程', which can appear in normal user requests unrelated to this skill. That can cause unintended activation and disclosure of internal execution details when the user did not explicitly request this behavior.
The module description and user-facing strings indicate the tracker is designed to output execution steps in Chinese, with no indication that users can choose another language or locale. This is a natural-language policy concern because it imposes a specific language by default rather than offering an opt-in or documented locale constraint.
This Python file contains user-facing status messages in Chinese for the enable, disable, and status commands. Because the skill does not offer a language or locale choice or document a justified locale restriction, this creates a natural-language policy issue under the language/locale rule.
The trigger '过程' is too short and generic, so it may activate on many benign inputs unintentionally. While this is primarily a trigger hygiene issue, in this skill it compounds the risk of unnecessary disclosure because activation enables verbose execution tracing.
No suspicious patterns detected.
