#Continue

I tested 7 AI coding agents in one afternoon. Only 2 finished.

Same React refactor. Same Sonnet 4.6 endpoint. Claude Code and Cursor one-shot it. Devin spent 87 minutes and broke unrelated tests. Here ar