Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only workflow helper with overly broad activation wording, but I found no hidden code, credential access, persistence, data exfiltration, or destructive behavior.
This skill is reasonable to install if you want a lightweight workflow/checklist helper, but its activation rules should be narrowed or implicit invocation disabled if your environment allows it, because it may otherwise appear during unrelated image, editing, or bug-fix requests.
The manifest description says to use the skill when a user asks for broad categories like work-productivity, practical workflow, artifact, checklist, analysis, or implementation support. Those conditions are common across many benign requests, so the description can cause ambiguous or accidental invocation that routes unrelated tasks through this skill.
The trigger keyword list includes extremely broad terms like "nano," "banana," "pro," "generate," "edit," and "bug fix," which can match many unrelated user requests. This creates overbroad skill activation, increasing the chance the skill is invoked outside its intended scope and influences responses in contexts the user did not intend.
The trigger list includes very broad everyday terms such as "nano", "banana", "pro", "generate", "edit", and "images", which can cause the skill to activate for many unrelated requests. Over-broad activation increases the chance that users are routed into an irrelevant workflow, producing unintended guidance or interfering with more appropriate skills.
The trigger sentences include very broad natural-language phrasing such as "Help me" and "I need a practical workflow," which are common across many unrelated requests. In an agent-routing system, this can cause unintended invocation of the skill, leading to misrouting, irrelevant workflow execution, and possible context leakage into a skill the user did not explicitly intend to use.
The invocation guidance lists generic keywords ('nano', 'banana', 'pro', 'generate', 'edit', 'images') and trigger sentences without clear scope boundaries or exclusion conditions. This increases the chance of accidental invocation for unrelated productivity, image, or general editing tasks, which can misroute agent behavior and reduce safety and predictability.
The trigger example uses a very broad everyday phrase ('Help me ...'), which can cause the skill to activate in many unrelated conversations. In an agent environment, overbroad activation can hijack routing, cause the wrong workflow to run, and lead to unintended handling of user requests outside the intended Nano Banana workflow scope.
The trigger phrases are broad enough to match common user requests such as asking for a practical workflow or help with bug fixing, which can cause the skill to activate outside its intended scope. In an agent environment, overbroad activation can route unrelated tasks through this skill, increasing the chance of unintended instruction injection, user confusion, or inappropriate workflow execution.
The example trigger phrases are vague, truncated, and resemble normal requests for help, which reinforces broad matching behavior rather than demonstrating clear, bounded activation. In systems that use examples to guide routing, this can further bias selection toward this skill for unrelated prompts.
该技能文件为 SKILL.zh-CN.md，全文以中文呈现，但未说明是否允许根据用户偏好切换语言，也未提供语言选择或 opt-in 机制。按规则，若技能自然语言说明隐含强制特定语言而没有用户选择，可能构成语言/区域策略问题。
The description says the skill should be used when a user asks for broad terms like work-productivity, nano, banana, or pro, or whenever they need a practical workflow, checklist, analysis, or implementation support for the requirement. This scope is too vague and expansive, making activation boundaries unclear and increasing the risk of accidental invocation on unrelated tasks.
The default prompt and skill metadata use very broad trigger terms such as 'work-productivity', 'practical workflow', 'artifact', 'checklist', and 'analysis', which can cause the skill to activate in many unrelated contexts. Because implicit invocation is enabled, this increases the chance of unintended routing, prompt injection exposure, or the skill being selected when a more appropriate and safer skill should handle the request.
The invocation guidance relies on broad phrasing and does not define clear boundaries for when the skill should or should not activate. This increases the chance that ordinary productivity, implementation, or workflow requests will spuriously match, which can degrade system reliability and may expose user context to an unnecessary skill path.
This Chinese README contains key descriptive and operational text in English, which can impose an unexpected language requirement on users of the localized document. There is no indication that users can choose their preferred language or that the mixed-language content is intentional and justified.
