Security audit
Security checks for vulnerabilities and agentic risk
This is a Markdown-only OpenAPI/Swagger documentation helper with no executable payload or persistence, though its automatic trigger wording is broader than ideal.
Before installing, consider whether automatic selection on broad API or developer-experience wording is acceptable. The skill itself is limited to documentation/workflow help and does not install code, persist state, or request sensitive access.
The trigger sentence is so broad and malformed that it can match ordinary user phrasing unrelated to explicit invocation of this skill. That increases the chance of unintended activation, causing the agent to route requests into this skill when the user did not clearly ask for it, which can lead to incorrect tool selection and prompt-scope confusion.
The trigger phrases are generic and formulaic enough that normal user requests about OpenAPI or API documentation could invoke the skill unintentionally. Unintended invocation can route user requests into the wrong workflow, causing inappropriate handling, misleading outputs, or expansion of the skill’s influence beyond what the user explicitly requested.
The trigger phrases are broad and partially generic, including natural-language patterns like 'Help me' and 'I need a practical workflow' tied to a long requirement sentence rather than a narrowly scoped invocation. This can cause unintended skill activation on unrelated user requests, leading to context hijacking or the skill being selected when another tool would be more appropriate.
The manifest description uses very broad routing language such as multiple generic domains and task types, which can cause this skill to be invoked for loosely related requests rather than only OpenAPI documentation work. Over-broad activation increases the chance the agent applies the wrong workflow, captures requests intended for other skills, or responds with lower-fidelity guidance in security- or architecture-sensitive contexts.
The trigger keywords include broad terms such as 'software-and-data' and 'developer experience' that are not specific to OpenAPI documentation tasks. This weakens activation boundaries and can cause unintended routing to this skill for many unrelated engineering requests, increasing misapplication risk and reducing trust in skill isolation.
The example trigger sentences are vague, truncated, and malformed, so they do not clearly define when the skill should activate. Ambiguous examples can train or signal permissive matching behavior, leading to accidental invocation and unpredictable skill selection in multi-skill environments.
The skill description and usage condition are broad enough to match many ordinary API or software-documentation conversations, which can cause the skill to activate outside its intended scope. Over-broad activation increases the chance of irrelevant instruction injection into unrelated user workflows and can degrade routing safety and predictability.
The keyword list includes generic terms like 'openapi', 'swagger', 'rest api', and 'developer experience' without qualification, making accidental invocation likely during normal technical discussion. In an agent system, this can cause unintended skill selection, leading to context confusion, unnecessary prompt expansion, and increased exposure to downstream instruction conflicts.
The default prompt is overly broad and partially malformed, which increases the chance this skill is invoked for generic backend or documentation-related requests rather than explicit OpenAPI tasks. In an implicitly invokable skill, vague routing language can cause unintended activation, leading to irrelevant guidance, context leakage into the skill, or reduced user control over which agent handles the request.
Enabling implicit invocation without tight activation constraints allows the skill to be selected automatically on loosely related requests. Because this skill is meant for documentation generation and validation, unintended invocation could expose user-provided repository, API, or design context to a skill they did not explicitly choose, and can amplify the risk created by the broad default prompt.
The trigger scope mixes broad keywords with malformed example phrases, lacking clear boundaries for when the skill should or should not activate. In an agent system, this can cause over-triggering on common software conversations, degrading routing integrity and potentially exposing downstream processing to irrelevant or unintended inputs.
This is presented as a Chinese README, but key user-facing content including the requirement summary, workflow description, keywords, and trigger phrases is written in English. That can amount to an implicit language constraint without user opt-in or explanation.
