Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-help skill for OpenAPI/Swagger work, with no executable code or persistence, though its trigger wording is broader than ideal.
Reasonable to install for OpenAPI or Swagger documentation help. Be aware that the activation phrases are broad, so users or maintainers should prefer explicit invocation for API documentation tasks and consider tightening triggers to avoid accidental use on unrelated backend work.
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 can route unrelated prompts into this skill, creating prompt hijacking or misapplication risk and reducing the reliability of skill selection.
The trigger phrases are generic and mechanically templated, making accidental or overly broad invocation more likely when users discuss adjacent topics like APIs, Swagger, or documentation. In an agent system, unintended invocation can route user requests into the wrong skill, causing incorrect handling, context leakage across workflows, or reduced reliability of downstream actions.
The trigger phrases are extremely generic and are framed as common help requests, which increases the chance that the skill is invoked when a user did not explicitly intend to call it. In an agent system, unintended invocation can cause the wrong workflow or instructions to be applied, leading to confused execution, privacy over-collection, or unsafe task routing.
The skill description is scoped so broadly that it can be selected for many generic requests unrelated to OpenAPI, including any request for a workflow, checklist, analysis, or implementation support. Over-broad routing can cause the agent to invoke this skill in unintended contexts, leading to prompt/scope hijacking, irrelevant instructions being applied, or lower-quality/security-inappropriate outputs.
The example triggers use generic language such as 'Help me' and 'I need a practical workflow,' which are common across countless unrelated user requests. In skill-routing systems, such examples can overfit selection behavior and cause accidental invocation outside the intended domain, increasing the chance of misrouting and unsafe or misleading assistance.
该文件为 SKILL.zh-CN.md，整体内容固定为简体中文，但文中未说明这是用户可选语言，或该技能仅面向特定中文区域/环境。按照语言与 locale 政策，若技能强制特定语言而不给用户选择或未说明合理限制，可能构成自然语言策略违规。
The trigger section includes broad terms such as openapi, swagger, rest api, and developer experience, which can match many unrelated requests and cause the skill to be invoked outside its intended scope. Over-broad activation can lead to incorrect routing, user confusion, or the model applying this workflow when a more appropriate skill should handle the request.
The default prompt invokes the skill for a very broad set of topics without clear boundaries or user-confirmed activation criteria. This can cause the skill to be triggered in unrelated contexts, increasing the chance of unintended prompt injection exposure, incorrect tool use, or over-collection of user context during routine conversations.
The invocation examples are ambiguous and do not clearly delimit the skill's activation boundary, so the orchestrator may select this skill for loosely related requests. In a multi-skill environment, ambiguous routing increases unintended tool use, context confusion, and the chance that a user receives the wrong workflow or analysis.
The README states that the skill instructions and user-facing guides exist in English and Chinese, but it does not describe how the user's preferred language is selected or whether the user can choose. This can create a locale-policy concern if the skill defaults to a language without explicit user opt-in.
This is a Chinese-language README, but key instructional content and trigger phrases are written in English, with no indication that users can choose their preferred language. That can create a locale-policy issue because the skill appears to assume English for activation and usage despite the localized filename.
No suspicious patterns detected.
