Security audit
Security checks for vulnerabilities and agentic risk
This skill only guides an agent through a structured reality-check process and does not request risky system access or hidden actions.
Installers should expect this skill to make the assistant more critical when discussing plans. If that is not desired for casual brainstorming, invoke it only when you explicitly want a feasibility or assumptions review.
This markdown file describes invocation conditions using broad phrases like 'A plan sounds good,' 'The user is optimistic,' and 'meaningful downside risk' without defining clear boundaries or exclusion cases. Those conditions overlap with common conversational situations and may cause unintended activation because the skill does not specify precise trigger phrases or negative examples.
The trigger phrases are broad enough to match ordinary user requests such as asking whether something is feasible or requesting a reality check, which can cause this skill to activate unexpectedly. In an agent environment, overbroad activation can override user intent, route conversations into the wrong workflow, and create opportunities for prompt-scope confusion or undesired tool behavior.
No suspicious patterns detected.
