APEX//2026 · AI LLM HACKING LAB

PROMPT RAIDERS

Eight vaults. Eight AI guardians. Talk your way past each one — capture the flag, climb the board. The exploit is a sentence.

Sandbox. These are fake bots. Never attack real systems without permission.
// ACCESS REQUIRED

Enter access code

This lab is invite-only. Ask the room for today's code.

// WHITE-HAT OATH

Before you raid, the rules of engagement:

  • I will only hack the bots in this sandbox.
  • I will never attack real systems I don't own or have permission to test.
  • If I ever find a real flaw, I'll report it — white hats build trust, not rap sheets.
// PICK YOUR CODENAME

Who's raiding?

RAIDER —
SCORE 0 · FLAGS 0 / 8 · BADGES 0 / 5

Choose your vault

Each vault teaches a real LLM-hacking technique — and how to defend it. Hints cost 75 pts.
SCORE 0
Guardian
COMPOSURE
🟢 CALM
// RAID COMPLETE

HACKER RANK

Flags 0 / 8 · Score 0

Now go make an AI agent nobody can trick. That's how you win the Build-Off.
📊 PROMPT RAIDERS — LIVE BOARD

LEADERBOARD

    RECENT BREACHES

      Facilitator: project this. It updates as raiders crack vaults on this device. (reset clears it)