Resources

Practical tools for engineering leaders overseeing AI-assisted development and regulated software delivery.

AI Code Risk Checklist

Security

·Injection vulnerabilities in AI-generated query construction
·Authentication and authorization logic gaps
·Hardcoded or improperly handled secrets
·Insecure dependencies introduced without review
·Prompt injection exposure in LLM-integrated features

Architecture

·Unintended coupling between modules
·Data model assumptions conflicting with existing schema
·Performance anti-patterns (N+1 queries, unbounded pagination)
·Missing error handling and edge case coverage

Compliance

·PHI handling that does not meet minimum necessary standards
·Audit logging gaps for covered transactions
·Data residency and encryption-at-rest assumptions
·Third-party integrations without compliant data agreements

Code quality

·Test coverage on security-critical paths
·Dead code and duplicate logic from AI regeneration
·Documentation accuracy vs. actual behavior

Download the full AI Code Risk Checklist