A code quality battle between 6 AI agents

I tested out 6 different AI agents - Loveable, Bolt, V0, Replit, Cursor, and Claude Code to see which would produce the best (and worst) quality code.

The task they were given was pretty simple and none of them generated complete garbage, but Claude and Bolt outperformed the others by a pretty decent margin.

Any thoughts on other quality metrics/methodologies I should use to compare?