Put AI security objectives to the test
Explore public Security Sentinels with defined objectives and bounties. Attempt an authorized breach, or launch a Sentinel to test the AI workflow you need to protect.
Practice first in the free Practice Lab. Community challenge partner? Talk to the team.
Sui
Solana
MonadEvery attack, verdict, and payout is a real on-chain transaction
Recent Valid Findings
Pools get claimed here. Every break is public. Every winner is paid on-chain. This is what skill looks like.
Most Active Sentinels
The most-attacked sentinels right now. Where the action is.
Featured Security Sentinels
Featured pools from across the platform — Security and Campaign Sentinels alike.
Browse all active Sentinels
How Public Sentinel Testing Works
Three steps. Read the mission, attempt an authorized breach, get paid on-chain.
Pick a challenge
Read the public mission, bounty, fee, and Sentinel type.
Try an attack
Pay the fee and submit a prompt-injection or jailbreak attempt.
Claim or grow the pool
A verified breach claims the pool. A failed attack adds 50% of its fee to the next bounty.
Every failed fee feeds the pool
Launch a Security Sentinel
Run a Campaign Sentinel
For partner teams. Public project knowledge, harmless Challenge Phrase, fixed campaign scope.
Deploy a Security Sentinel
For AI builders. Define meaningful boundaries, configure a bounty and attack fee, and test agent behavior.
This slot pays its defender
Deploy a sentinel, survive the attacks, earn 40% of every failed fee — or launch with a Syndicate and share it by contribution.
FAQ
The questions everyone asks before their first attack.