Security audit
Security checks for vulnerabilities and agentic risk
This Markdown-only skill gives a disclosed Dunning-Kruger coaching framework, with the main caution that its broad triggers could over-label ordinary confidence as bias.
Use this skill for concrete confidence claims where external evidence can be checked. Avoid relying on it as the sole basis for personnel, medical, financial, or other high-stakes judgments, because its broad cues can frame people as overconfident too readily.
The activation criteria are broad and subjective, relying on cues like perceived novice status, confidence mismatch, or absent feedback loops. In an autonomous agent, this can cause over-triggering on ordinary conversations or interpersonal judgments, leading the agent to apply a bias-diagnosis frame without sufficient evidence and potentially produce unfair or intrusive assessments.
The 'When to Use' guidance includes vague invocation cues such as mentions of being 'overconfident,' 'doesn't know what they don't know,' or even 'imposter syndrome,' which are common conversational concepts and not reliable security-grade triggers. This increases the chance that the skill activates on weak signals and encourages the agent to infer competence deficits from casual language, creating risk of misclassification and harmful downstream decisions.
No suspicious patterns detected.
