Security audit
Security checks for vulnerabilities and agentic risk
This documentation-only skill does not appear malicious, but its auto-activation scope is broad enough to steer unrelated user requests.
Review this skill before installing if your agent auto-loads skills. It should be narrowed to explicit SkillScan or ClawHub security-workflow requests, and implicit invocation should be disabled or constrained so ordinary words like security, must, before, or bug fix do not route unrelated work through this skill.
The trigger keywords and example invocations are broad enough to match ordinary requests such as generic security, workflow, bug-fix, or activation help. In an agent skill-routing system, this can cause unintended auto-selection of the skill, increasing the chance that unrelated user prompts are captured and processed under the wrong workflow, which is a real security and reliability boundary issue.
The skill description says to use the skill when a user asks for terms like "security," "gate," or "every," which are common words that can appear in many unrelated conversations. This creates overbroad auto-activation risk: the skill may be invoked outside its intended context, causing unintended workflow steering, prompt injection surface expansion, or interference with other skills.
The keyword list includes highly ambiguous words such as "security," "gate," "every," "must," "pass," "before," "activate," and "bug fix" without any scope constraints. In a shared agent environment, these common terms can match ordinary requests and cause this skill to activate excessively, potentially hijacking routing decisions and inserting unrelated instructions into benign tasks.
The trigger keywords include very common terms such as "security", "gate", "every", "must", "pass", and "before", which creates a high risk of accidental activation in unrelated conversations. Over-broad activation can route users into the wrong workflow, causing security guidance to be applied out of context or enabling prompt-surface expansion where the skill intercepts requests it was not intended to handle.
The trigger sentence is constructed from generic, everyday wording and can plausibly match ordinary user requests that were not intended to invoke this skill. In an agent ecosystem, overly broad activation increases the chance of unintended execution paths, causing the skill to handle prompts outside its intended security-review scope and potentially interfering with safer routing or user intent.
The keyword list includes extremely generic terms such as 'security', 'gate', 'every', 'must', 'pass', 'before', and 'activate', which are likely to appear in many unrelated conversations. In an agent ecosystem, such ambiguous activation cues can cause over-triggering, leading to incorrect workflow selection, confusion, and possible interference with security-sensitive tasks.
The trigger sentences are broad, natural-language phrases that can match ordinary user requests rather than a narrowly defined invocation pattern. This increases the chance of unintended skill activation, causing the agent to enter this workflow when the user did not explicitly request it, which can misroute tasks or override more appropriate skills.
This file is labeled as a Simplified Chinese README, but key user-facing content such as the title and requirement description is presented in English. That creates an implicit language/locale mismatch without giving the user an explicit choice or explaining that mixed-language output is expected.
The example trigger sentences reinforce vague invocation behavior by showing activation on broad, reusable wording rather than clearly bounded intents. This makes accidental or adversarial invocation easier, because attackers or normal users can include generic phrasing that appears natural while still forcing the skill into contexts where it does not belong.
The activation description says to use the skill when a user asks for broad topics like work-productivity, skillscan, security, gate, or every, but it does not define boundaries or non-matching cases. This ambiguity increases unintended triggering and makes it harder for the orchestrator to distinguish this skill from other generic productivity or security skills.
The example triggers are repetitive and only show positive matches, with no counterexamples to constrain use. Without representative positive and negative examples, broad phrases can be overfit by the router and cause false activations in unrelated requests.
The default prompt is broad and generic enough to match ordinary user requests unrelated to a narrowly scoped security workflow. This increases the chance of accidental or overly broad invocation, causing the skill to inject instructions or context into conversations where the user did not clearly request it.
Enabling implicit invocation without strong trigger constraints allows the platform to auto-select this skill for loosely related requests. In combination with the broad prompt and expansive description, this can lead to unintended activation, prompt-scope creep, and unreviewed influence over user interactions.
The usage signals list broad keywords such as 'security', 'gate', 'every', and 'bug fix' without sufficient boundaries, making the activation scope ambiguous. This can cause the skill to trigger in many unrelated contexts, leading to prompt hijacking of normal conversations, incorrect tool selection, and accidental execution of a workflow the user did not request.
