hunt-agent-guardrail-evasion

SUSPICIOUS. This is an explicitly dual-use guardrail-evasion skill: it provides actionable techniques and a probe intended to characterize and bypass safety/DLP controls. No evidence shows credential theft, exfiltration to an attacker-controlled host, hidden load-time execution, or an installer chain; the example endpoint and cookie are placeholders. Risk stems from the capability to defeat safeguards and potentially unlock harmful downstream actions, not from confirmed malware.