secure-code-review

High-risk but not malware. The skill is internally consistent as a security-audit/offensive-review workflow, and installer sources are mostly official, but it meaningfully expands an AI agent into offensive security operations, broad secret discovery, local script execution, and transitive trust in other skills.