Audit for unauthorized access -- authentication, unprotected routes, IDOR, input validation, and role-based permissions.
Audit for cost explosion risks. Checks AI provider spending limits, hosting caps, per-user rate limits, and spam protection. Redirects to provider dashboards when settings can't be verified from code.
Audit for legal risks before charging users. Checks privacy policy, AI tool code ownership terms by plan, and copyleft license dependencies.
Audit for blind spots in monitoring and recovery. Checks error handling, error monitoring, uptime alerts, database backups, and version control. Asks questions when settings can't be verified from code.
Audit for exposed secrets, open databases, and data over-exposure in client-side code.