Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-helper skill for OpenAPI and Swagger work, with some overly broad activation wording but no executable code, persistence, credential handling, or hidden behavior.
This skill is reasonable to install for OpenAPI or Swagger documentation help. Be aware it may activate on broad API or developer-experience phrasing; invoke it explicitly when you want it, and avoid pasting secrets into API examples unless you have redacted them.
The trigger phrases are generic and expansive, which can cause the skill to activate in contexts broader than intended. In an agent system, over-broad activation can route unrelated user requests into this skill, producing inappropriate guidance, increasing prompt-surface exposure, and creating opportunities for instruction interference across skills.
The trigger phrases are generic and malformed enough to match loosely phrased user requests, which increases the chance this skill is invoked when the user did not explicitly intend to use it. In an agent ecosystem, unintended invocation can cause misrouting, disclosure of irrelevant context to the skill, or bypass of a more appropriate and safer workflow.
Although this is the Chinese README, substantial user-facing content including the title, demand description, workflow description, keywords, and trigger phrases is presented in English. This can effectively impose a language expectation on users without explicit opt-in or explanation.
The skill description and activation guidance are broad enough to match many ordinary requests involving APIs, documentation, or software/data work, which can cause over-triggering. In an agent environment, this can lead to the wrong skill being invoked, producing irrelevant guidance, displacing more appropriate skills, or expanding the attack surface for prompt-routing abuse.
The keyword list contains generic terms such as 'software-and-data' and broad API-related phrases without activation constraints, making accidental or adversarial triggering easier. Overly generic triggers can be exploited to steer unrelated conversations into this skill, reducing routing precision and potentially interfering with safer or more relevant behaviors.
The example trigger sentences use vague, natural phrasing and effectively suggest activation on loosely related requests, without clarifying what should not trigger the skill. This increases ambiguity in skill selection and makes prompt-injection-by-routing easier because ordinary user language can unintentionally activate the skill.
The skill description and usage conditions are broad enough that it could be invoked for loosely related requests, causing incorrect routing or over-application of the skill. In an agent system, overly permissive triggers can lead to context confusion, irrelevant instructions being injected into conversations, and reduced reliability of downstream actions.
The keyword trigger list is ambiguous and lacks scope constraints, making accidental activation likely when users mention generic API or developer-experience topics. This is dangerous in multi-skill environments because it can cause the wrong skill to take control, produce irrelevant guidance, or suppress a better-matched specialized workflow.
The skill enables implicit invocation (allow_implicit_invocation: true) but the metadata does not define narrow activation constraints, scoped triggers, or guardrails limiting when the skill should be selected. This can cause the agent to invoke the skill in broader contexts than intended, increasing the chance of unintended prompt/data exposure or unsafe action chaining based on loosely related user requests.
The activation examples are underspecified and ambiguous, so they do not clearly constrain when the skill should run. In an agent environment, ambiguous routing conditions can cause unintended skill invocation, enabling context confusion and making it easier for unrelated requests to be handled by a skill with mismatched assumptions or instructions.
The trigger sentence is broad enough to match ordinary user phrasing about getting help, which can cause the skill to activate outside its intended scope. Over-broad activation increases the chance of prompt routing mistakes, irrelevant instruction injection into unrelated conversations, and accidental exposure of this skill's behavior in contexts where it was not requested.
The file lists separate English and Chinese instruction/readme files, but it does not explicitly state that the user can choose their preferred language or locale. Because language handling is described in natural language, this may create ambiguity around whether a specific language is imposed without opt-in.
该文件整体为 zh-CN 中文版本，但文本中未说明这是可选语言呈现，亦未给出用户语言选择机制或区域性约束理由。按文件内容看，技能可能默认以特定语言/区域输出，与语言/locale 需经用户选择或明确正当化的政策不符。
No suspicious patterns detected.
