Establishing collaboration across LLM Agents when context is close to saturation | Hacker News Reader