Prime Agent: A Self-Improving RLM Agent | Dark Hacker News