bb-local-toolkit

This is not coherent as a low-risk helper skill: it is an offensive-security playbook that enables an AI agent to scan, probe, and validate exploits on external targets, including AI-specific prompt extraction and CI/CD attack chains. The install source is only mildly risky, but the core capability set, OOB exfiltration endpoints, and agentic exploit guidance make the skill high risk and suspicious rather than benign.