Hands-On Guide to Testing a Service From Outside: Setup and First Run

This is part four of a series on self-defense security for services in the age of AI, shifting from theory to a hands-on demonstration. The authors ran an autonomous legitimate scanner and an unrestricted attacker-grade tool against a copy of their own service, never touching production. A cheap AI model driving a legitimate scanner, limited to four login routes, found three weaknesses in the audit and session layer, costing about $0.45. The article explains that AI tools help defenders cover the 231 exposed entry points far faster than manual review. The next installment will compare what the attack tool and two outside scanners found.