Tag
A minimal 2-step LLM chain that implements a planner + coder pipeline, designed to fit on a single GPU without requiring a full agent framework.