Add prompt-defense-audit to Code section

[ppcvote]

Adds prompt-defense-audit — a static 12-vector defense posture scanner for AI system prompts.

• Pure regex, <5ms, zero dependencies, MIT license
• Checks for missing defenses: prompt injection, data leakage, role escape, indirect injection, output weaponization, and 7 more vectors
• npm package: npx prompt-defense-audit "Your system prompt"
• Merged into Cisco AI Defense mcp-scanner
• Proposed to Microsoft Agent Governance Toolkit (assigned by maintainer)

[ppcvote]

Friendly ping — any chance to review this? Happy to make changes if needed.