oxidized-agentic-audit 0.5.0

Security scanning for AI agent skills — scans skill directories for dangerous bash patterns, prompt injection, supply chain risks, secret leakage, and frontmatter quality issues