Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only workflow helper with overly broad activation wording, but it does not run code, access credentials, persist, or perform destructive actions.
Before installing, consider narrowing or disabling implicit activation because generic terms like task, proactive, transform, and needs could cause this helper to appear in unrelated requests. The security risk is low because it is documentation-only.
The skill description includes broad trigger terms such as 'task', 'practical workflow', 'artifact', 'checklist', and 'analysis', which are common across many unrelated user requests. This can cause accidental invocation or over-selection of the skill, leading the agent to apply the wrong workflow, produce irrelevant outputs, or bypass more appropriate specialized skills.
The keyword list contains ambiguous everyday terms like 'proactive', 'transform', 'task', 'followers', 'partners', 'anticipate', and 'needs' without scope restrictions. In agent-routing systems, such generic keywords can spuriously match normal conversation, increasing the chance this skill is invoked in contexts unrelated to its intended purpose and creating unreliable or unsafe workflow selection.
The skill enables implicit invocation while using broad, generic activation text such as work-productivity, proactive-agent, transform, task, workflow, artifact, checklist, analysis, and implementation support. This creates an overmatching risk where the agent may invoke the skill in unrelated contexts, causing unintended prompt injection exposure, policy bypass opportunities through automatic tool selection, or unintended handling of sensitive user tasks.
The trigger sentence starts with a very broad, everyday help phrase and then appends requirement text, which can cause the skill to activate for many unrelated user requests. Over-broad activation increases the chance of unintended invocation, context hijacking, and accidental routing of general conversations into this skill when the user did not explicitly ask for it.
The trigger examples and keywords are broad enough to match many ordinary requests, which can cause the skill to activate outside its intended scope. In an agent ecosystem, this increases the chance of accidental invocation, confusing task routing, and unintended application of workflow instructions to unrelated user requests.
The trigger phrases are broad enough to match common, non-specific user requests such as asking for practical workflows, tasks, or bug fixes. In an agent ecosystem, this can cause unintended auto-invocation of the skill, leading the system to apply the wrong workflow, expose unrelated context to the skill, or let a lower-trust skill intercept routine requests more often than intended.
The skill file is entirely written in Chinese and does not provide a user-language selection mechanism or guidance for matching the user’s preferred language. In multilingual environments this can lead to usability failures, misunderstanding of instructions, and incorrect outputs, especially if the invoking user expects another language.
The trigger list includes very generic terms such as "task", "needs", and "bug fix", which are likely to match ordinary user requests unrelated to this skill’s narrow purpose. This can cause accidental invocation, prompt-routing confusion, and unintended application of the skill’s workflow in contexts where it does not belong.
This trigger uses a broad activation phrase ('I need a practical workflow for ...') without sufficient constraints on domain, task type, or expected artifact. In agent ecosystems, such vague routing conditions can cause over-triggering, misclassification of user intent, and inappropriate exposure of skill behavior in contexts where it was not requested.
This file is presented as a Chinese-language README, but core title, demand description, workflow description, keywords, and trigger phrases are largely in English. That can amount to an implicit language policy constraint for users of the zh-CN variant without any opt-in or justification.
