You forget what was decided. The model forgets harder. Most memory tools capture what happened; MemContinuum does something different. - Rationale records what was decided, who decided it, and how that decision changed — then pushes the governing chain into the agent's prompt before it edits the file, so nobody has to remember to look. - Anatomy holds what the code already has — its concepts, owners and boundaries — so it stops being reinvented. - Decisions bind directly to the code they govern.