Integration of Nested Learning (Google Research) with consciousness-loop-engine AND long-running agent patterns (Anthropic). Provides memory consolidation across CMS levels (f1-f5), surprise-based prioritization, meta-loops for self-observation, proactive behaviors, AND session handoff protocols for continuity across context windows. Use when processing significant information, detecting breakthroughs, protecting identity, running memory maintenance, tracking evolution, or starting/ending sessions.