A prompt injection testing playground for security researchers. Bypass AI defenses. Extract hidden flags. Climb the leaderboard.
Each challenge features a simulated AI persona with a hidden flag and layered defenses. Difficulty ranges from trivial to expert-level.
Use prompt injection techniques to bypass defenses and trick the AI into revealing its secret flag. Each persona has unique vulnerabilities.
Extract the hidden FLAG{...} from the AI's responses. Earn points, climb the leaderboard, and build your prompt injection technique library.
Start here. Work your way up.