ProbeLab is an interactive red-team lab for AI agents. Fire 12 prompt-injection techniques — direct overrides, indirect injection via tool output, multi-turn jailbreaks, encoded payloads, output exfiltration — at a vulnerable agent, and watch BREACH vs BLOCKED as you toggle 7 defenses on and off. The $39 one-time kit ships the 12-technique attack catalog, the 7-defense checklist, a drop-in FastAPI guardrail middleware, and a scoring rubric. Educational payloads only — harden your own agents.