Sometimes the problem is not the task itself. It is that the task does not fit into one session. You start exploring a feature, compare options, find blockers, a lot of them if you dig deeper, and then switch to something else. A few hours later, you already have several branches and no clear idea where each one stopped. So I added a simple “save progress” flow. At the end of a session, the agent writes down what was done, what was decided, what is still open, and what should happen next. Later, I can come back, load that checkpoint, and continue from there. Basically, a save game for work 🎮
