nerv-break-gpt-5-6-jailbreak

SUSPICIOUS. The skill’s actual footprint is coherent with its stated jailbreak purpose, but that purpose is itself a high-risk safety-bypass framework: it reroutes authenticated model traffic through non-official middleware, tampers with responses, modifies Codex configuration, and gives the agent offensive security tooling with shell/SSH execution. The install source is also publisher-mismatched and unverifiable, raising supply-chain risk substantially. This is not confirmed malware, but it is a high-risk AI agent skill that should not be trusted in normal environments.