Benchmark Optimization Loop

This turns "make it faster" requests into disciplined benchmark loops with correctness gates and variant tracking. It won't let you optimize until you have a baseline metric, a way to verify correctness, and a search budget. The loop is simple: measure, hypothesize, test variants one at a time, reject anything that breaks or can't reproduce, promote the winner. The variant table format and promotion gates keep you honest about what actually improved and why. Use this when someone asks for 20x speedups or recursive optimization and you need to convert enthusiasm into a bounded, measurable process that won't break production or burn time chasing noise.

Install to Claude Code

npx -y skills add affaan-m/everything-claude-code --skill benchmark-optimization-loop --agent claude-code