Lifecycle Management is the discipline of treating memory as an active operating loop rather than a passive database. A robust shared brain does not wait for the model to decide when to search or write memory. Instead, it defines lifecycle stages around every task: pre-fetch relevant history before reasoning starts, post-turn capture decisions and file modifications after the agent acts, and pre-compress the session before the context window truncates away useful state. This sits at the junction of L2 (Knowledge & Memory) and L4 (Orchestration) because it coordinates both retrieval and control flow.
clawmemlanggraphgsd