Run autonomous, metric-driven experiments on a version-controlled implementation against a fixed evaluation harness. Use when the user asks to improve eval pass rate, benchmark score, prompt or policy quality, performance, cost, or another measurable outcome through repeated hypothesis, change, eval
Run autonomous, metric-driven experiments on a version-controlled implementation against a fixed evaluation harness. Use when the user asks to improve eval pass rate, benchmark score, prompt or policy quality, performance, cost, or another measurable outcome through repeated hypothesis, change, evaluate, keep-or-discard cycles. Protect generalization with holdout gates and reject hardcoded cases,