0G Sentinel

0G Sentinel is the first on-chain AI agent security layer built natively on the 0G stack.

AI agents are already moving real money across DeFi protocol but there is no way to know if an agent's code is safe, whether it has been audited, or whether its on-chain

behavior is trustworthy. We lost over $88 million to AI agent exploits in 2024–25, and the attack surface is only growing.

0G Sentinel fixes this with three primitives. First, a dual AI scan pipeline: behavioral analysis of the agent's full transaction history, and a static Solidity vulnerability scan of its code, both powered by 0G Compute. Second, an AttestationRegistry contract on 0G Aristotle Mainnet that writes the risk score immutably on-chain so anyone can verify it at any time. Third, AgentGate, a composable security contract that any DeFi protocol can drop in with a single require() call to block unscanned or flagged agents before they execute.

The full pipeline is live. Submit any agent address, get a 0–100 risk score across four tiers, and have the result attested on-chain, all in one call. Other AI agents can query 0G Sentinel directly via the OpenClaw skill registry. DeFi protocols integrate via a single REST call or the AgentGate contract. Three contracts are deployed to 0G Aristotle Mainnet (Chain ID 16661), and the project is fully open source.