AI

Don’t trust the LLM. It’s not deterministic. And while the capabilities of coding agents such as Claude Code CLI are impressive, you shouldn’t mistake them for an actually trust-worthy entity. That’s why we look at how to put our coding agent into a cage, making it a: Cagent! … Anyone?

The internet is divided, one half has gone full-blown Coding Agent Mode, adamant not to touch any code ever again and the other half is skeptic AF and trying to convince everyone that AI generated code still mostly sucks. So, this is just me trying to see for myself and sharing my experience after my first actually successful result from Vibe Coding or Agentic Engineering or whatever.