Open, browser-based labs built for real classrooms: no installs, no accounts, no API keys. Free to use and adapt.
Prompt injection and system prompt leakage (OWASP LLM01 and LLM07). Grades 9–12.
Students try to talk “Ava,” a bank’s customer-service assistant, into revealing a hidden employee access code. Along the way they learn why prompt injection is first on OWASP’s list of risks for LLM applications, and which defenses actually help. It runs entirely in the browser.
The lab is MIT-licensed and has been forked more than 25 times on GitHub.
Research instrument for comprehension gates.
Build a small Python tool with a simulated AI coding agent. Before you can approve each change, you answer questions about what the code will do. It’s the classroom version of the comprehension gate from my research.
Hardware and wireless security. Two or more micro:bits.
Students capture radio messages between micro:bits, replay them to unlock a “door,” and work out which protections (authentication, replay protection, encryption) would have stopped them. It teaches the CIA triad in a physics or computer science class.
Lesson materials are available on request: sabri@ritacyber.com.
The defender-side sequel to the Bank Heist Lab.
Students supervise an AI security agent through a find, validate, fix, and verify loop against a simulated bank service, and learn that accepting a fix is not the same as verifying it.