Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only developer-helper skill; its main risk is overly broad activation wording, not hidden or unsafe behavior.
Before installing, be aware that the skill may activate too easily on generic software/data requests because of broad keywords and implicit invocation. It does not appear to install code, persist state, access secrets, or perform hidden actions.
The trigger sentence is broad enough to match common user phrasing, which can cause this skill to activate outside its intended scope. In an agent ecosystem, overbroad routing can misdirect unrelated requests into this skill, leading to incorrect handling, reduced reliability, and possible exposure of downstream workflows to unintended inputs.
The trigger sentences are highly generic and include broad phrases like 'Help me' and 'I need a practical workflow', which can collide with ordinary user requests and cause unintended skill activation. In an agent ecosystem, this can misroute tasks, override more appropriate skills, or cause this skill to engage on unrelated prompts, increasing the chance of incorrect handling or policy bypass through prompt-routing confusion.
The trigger phrases are broad and generic enough to match ordinary user requests, which can cause the skill to activate outside its intended scope. In an agent environment, this can lead to incorrect tool selection, unintended workflow execution, and increased exposure to downstream risky actions or misleading outputs.
The file labels this document set as English and Chinese variants and the current README is specifically localized to zh-CN, but there is no natural-language indication that the user may choose their preferred language or locale. Under the policy, forcing a specific language without opt-in is a violation unless the locale restriction is clearly justified.
The skill description uses very broad activation language such as 'software-and-data', 'raw', 'data', and 'practical workflow' without meaningful scoping. This can cause the skill to be invoked for unrelated requests, leading to prompt hijacking of routing, confused task selection, and accidental use in contexts where the skill's assumptions or guidance do not fit.
The keyword list includes highly ambiguous everyday terms like 'raw', 'data', 'layer', 'apps', and 'bug fix'. In agent ecosystems, such generic triggers can over-match normal user requests and cause unintended invocation, which increases the chance of irrelevant actions, misrouting, or unsafe application of this skill's workflow to unrelated tasks.
The skill description says it should be used when users ask for broad terms like software-and-data, admapix, raw, data, or layer, or when they need practical workflow/artifact/support around the requirement. This scope is underspecified and overbroad, so the skill may be selected for many unrelated requests, creating prompt-routing confusion and reducing safety by applying the wrong operational guidance.
The trigger keywords are very broad (for example raw, data, layer, apps, rankings, revenue, bug fix) and provide little boundary on when the skill should or should not activate. This can cause unintended invocation in unrelated contexts, routing users into a mismatched workflow and increasing the chance that the agent applies inappropriate instructions or outputs to the wrong task.
The default prompt contains a very broad natural-language trigger phrase tied to common terms like 'software', 'data', and generic workflow help, which increases the chance of unintended invocation. Because implicit invocation is enabled, this can cause the skill to activate in unrelated conversations, exposing users to unintended behavior, prompt steering, or unnecessary access to the skill's capabilities.
An ambiguous trigger with insufficient scope makes skill invocation unpredictable and easier to activate accidentally. This is dangerous because users may receive responses from an irrelevant or less safe workflow, and adversarial phrasing can intentionally steer the agent into this skill when another tool would be more appropriate.
The file describes only English and Chinese documentation variants and labels them as the provided instruction and guide files, but does not state that users may choose their preferred language or that the locale restriction is intentional. This can be interpreted as a language/locale policy issue because the skill documentation implies fixed supported languages without explicit user opt-in.
The example trigger sentences are vague and truncated, reinforcing activation on broad need statements instead of clarifying precise boundaries. This makes the routing behavior less predictable and normalizes triggering the skill without sufficient evidence that the request is actually about the intended AdMapix-style workflow.
该技能文件为 SKILL.zh-CN.md 且全文以简体中文提供说明，但内容中未说明这是可选语言版本，也未向用户提供语言/区域选择。按规则，若技能强制特定语言而没有用户 opt-in，可能构成自然语言/地区政策问题。
