execute-task

SUSPICIOUS. The skill is broadly consistent with a task-execution orchestrator, but its autonomous design, broad tool scope, subagent dispatch, and external-research pathways create meaningful security risk disproportionate to ordinary helper skills. I found no direct credential theft or malware behavior in this skill body, but it should be treated as high-trust automation.