TL;DR Autonomous coding agents are good at writing code. They are bad at knowing what's actually risky about the code they just wrote. I built Verishttps://github.com/vighriday/Veris -- an MCP-native verification intelligence layer. You point it a...