evaluating LLM coding agents in real-time strategy games

A benchmark leaderboard is research infrastructure, not a product a solo dev can monetize — the value is the frontier-model access and evaluation harness, not anything forkable.