octopus-security-audit

SUSPICIOUS: the skill is internally aligned with a security-audit purpose, but it grants an AI agent offensive security and semi-autonomous remediation capabilities, inspects broad local/repo history data, and funnels work through an external orchestrator/model stack. This is not confirmed malware, but it is a high-risk security skill that should be tightly sandboxed and require explicit user approval for exploit, patch, and validation actions.