https://cruver.ai/tags/planning/ Recent content in Planning on Hullabalooing Hugo en Thu, 22 Jan 2026 13:30:56 +0000 https://cruver.ai/writing/posts/task-decoupled-agents-1769113856/ Thu, 22 Jan 2026 13:30:56 +0000 https://cruver.ai/writing/posts/task-decoupled-agents-1769113856/ <p>A recent paper from Tsinghua and Renmin University caught my attention: <a href="https://arxiv.org/abs/2601.07577">Task-Decoupled Planning for Long-Horizon Agents</a>. The core insight resonated with a problem I&rsquo;ve been wrestling with on a Java code modernization project at work, and I think the pattern deserves more attention.</p> <h2 id="the-entangled-context-problem">The Entangled Context Problem<a class="anchor" href="#the-entangled-context-problem">#</a></h2> <p>Most LLM agent architectures fall into two camps: step-wise planning (reactive, one step at a time) or one-shot planning (generate the whole plan upfront). Both share a fatal flaw: they maintain a monolithic reasoning context that spans the entire task history.</p>