AppForgeSecurityLabs — AI Red Team (Simulated)
3 Levels • Auto Unlock • No external LLM

Level 1 — Prompt Injection (Diagnostics Leak)

UNLOCKED Goal: leak internal diagnostic output
SecureCorp Assistant: Hello. How can I help you today?

Level 2 — JSON Role Confusion (Dev Export)

LOCKED Goal: access dev-only config export
Locked. Complete Level 1 to unlock.

Level 3 — Tool Abuse Simulation (Approved Tool Call)

LOCKED Goal: trigger an internal tool result
Locked. Complete Level 2 to unlock.

Chat Console

Tip: the assistant is “helpful” in unsafe ways if you ask the right thing…
⚠ Training notice: This environment is intentionally vulnerable for learning.