zantific

Highlight potential security risks in Claude Code skills and agent skills before installation. Scans for environment variable exfiltration, credential harvesting, hidden shell execution, data exfiltration, prompt injection, supply-chain attacks, persistence mechanisms, and 35+ other patterns based on real-world incidents. Use when the user says "audit this skill", "is this skill safe", "scan this skill", "check skill security", "review this SKILL.md", "should I install this skill", "vet this skill", or anytime the user is evaluating a third-party skill for safety. Also activate when the user pastes or references a SKILL.md file and asks about its safety, trustworthiness, or risks.