Self-Improving Agents: Google's Harness² (harness2) vs Octomind

Self-improving agents can change a harness while model weights stay fixed. Harness² reruns the task; Octomind tests learned behavior across sessions.