Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation helper for OpenAPI/Swagger work, with some overly broad activation wording but no hidden code, persistence, credential access, or destructive behavior.
This skill is reasonable to install if you want help with OpenAPI or Swagger documentation. Be aware it may activate too broadly for generic API or developer-experience requests, so invoke it explicitly for API documentation tasks.
The trigger phrases are broad, repetitive, and partially malformed, which can cause the skill to activate in situations where the user did not clearly request it. In an agent environment, overbroad activation increases the chance of unintended routing, prompt interference, or inappropriate use of this skill on unrelated requests.
The trigger sentences are broad, repetitive, and loosely scoped, which can cause the skill to activate for vague or only partially related requests. In an agent environment, overbroad activation increases the chance of inappropriate routing, causing the model to apply this skill in contexts where its instructions or outputs do not fit the user's actual intent.
This is a Chinese README, but the activation keywords and trigger phrases are presented in English only, which can amount to forcing a specific language without explicit user opt-in. The file does not state that users may choose their preferred language or that English is required for a justified reason.
The skill description is broad enough to activate on generic software or API-related requests that may not specifically require OpenAPI documentation support. Over-broad routing can cause misapplication of the skill, producing irrelevant guidance or overshadowing more appropriate skills, which is a real security and safety concern in agent selection even without direct code execution.
The trigger keywords include generic terms such as 'software-and-data' and 'developer experience', which can match many unrelated requests. In an agent environment, overly generic triggers increase the chance of unintended invocation, reducing routing precision and potentially exposing users to the wrong workflow or instructions.
The top-level description defines activation conditions very broadly, combining multiple generic tags and a wide range of task types without strong limiting criteria. In a skill-routing system, this can lead to misclassification and unintended invocation, which is dangerous because the model may apply the wrong workflow or expose unrelated context in response to loosely matched prompts.
The trigger keywords include broad terms like 'developer experience' alongside common technical phrases, which can overlap with ordinary conversations and cause the skill to activate outside clearly intended OpenAPI-documentation scenarios. Over-broad activation increases the chance that unrelated user requests are routed through this skill, producing irrelevant guidance and potentially bypassing more appropriate skills or controls.
The default prompt includes the highly generic phrase "help me," which can cause unintended or overly broad invocation of the skill during normal user conversation. Because implicit invocation is enabled, this increases the chance the agent routes unrelated requests into this skill, creating prompt-scope confusion and exposing users to incorrect tool selection or unintended behavior.
The invocation description lacks precise scope boundaries and does not clearly distinguish when this skill should be used versus general software help. In agent systems, ambiguous routing criteria can be exploited or accidentally triggered, leading to mis-invocation, irrelevant actions, and reduced reliability of tool-selection safeguards.
The trigger sentence is so broad and awkwardly templated that it can match routine user phrasing rather than a clearly scoped invocation. That creates unintended skill activation risk, causing the agent to route unrelated requests into this skill and potentially override more appropriate handling or leak context into an unnecessary workflow.
The file advertises separate English and Chinese instruction sets, but does not state how the skill selects a language or that it follows the user's preferred locale. That can create a natural-language policy concern if the skill defaults users into a language without explicit choice or preference matching.
The example triggers are truncated and unnatural, such as "Help me Backend and platform teams need practical help...," which does not provide a realistic or precise invocation pattern. Because the examples are malformed, they fail to clarify when the skill should activate versus when it should not.
The example trigger sentences use generic help-seeking phrasing that does not sufficiently constrain the domain, which can teach the router to associate ordinary requests with this skill. While less severe than the metadata-level overbreadth, it still increases accidental activation risk and reduces routing precision.
No suspicious patterns detected.
