Security audit
Security checks for vulnerabilities and agentic risk
The skill does not install or run code, but it is explicitly designed to help rewrite AI text so it can bypass AI detectors.
Install only if you are comfortable with a skill whose main function is detector evasion. It should not be used to hide AI authorship, misrepresent originality, or bypass academic, workplace, or platform integrity checks.
The skill description openly promotes rewriting text to bypass GPTZero, Turnitin, and Originality.ai, but provides no warning, restriction, or policy boundary around that capability. This increases the likelihood of deceptive academic, compliance, or trust-and-safety misuse by normalizing evasion as the expected use case.
The 'Bypass' mode gives concrete, stepwise tactics to rewrite AI-generated content from scratch, inject human-like imperfections, and vary rhythm specifically to pass stricter detectors. This is actionable evasion guidance that lowers the barrier to fraudulent use, including academic cheating, policy circumvention, and misrepresentation of authorship.
The skill is explicitly designed to help users paraphrase AI-generated text so it appears human-written and avoids detection, which is a direct facilitation of deceptive behavior. Although it uses natural-language guidance rather than code, the operational goal is still evasion of integrity controls, making the context materially dangerous.
The trigger phrases are intentionally broad and include common rewriting requests such as paraphrasing and rewording, which can cause the skill to activate outside a narrowly controlled context. In this skill, that broad routing is especially risky because the skill’s stated purpose is helping users evade AI-content detectors, so ordinary editing requests may be funneled into misuse-oriented behavior.
No suspicious patterns detected.
