Autonomous agents are entering Web3 and finance fast. But what happens when a malicious prompt makes an agent sign a transaction or move funds? Could agent security become Web3’s next major security crisis? I wrote the full deep dive.