Show HN: AI vs. AI – code and reviews only count if they survive an attack

A developer introduces Pantheon, a technique where AI subagents generate, break, and rewrite code until it survives its own attacks. Pantheon-X pits models like GPT against each other, while Pancheon Gap checks the validity of reviews to derive surviving solutions. The project is open to feedback via GitHub issues.