memory-compounding-audit

Memory compounding occurs when captured observations and session summaries actively change behavior in later sessions. It requires classifying captured knowledge into compounding, dead, never captured, and noisy buckets to distinguish useful signal from stale context.

Designing retention and decay rules involves analyzing memory paths from capture points like hooks and external stores. You produce concrete promotion and resurfacing rules so repeated lessons become durable guidance instead of forgotten observations.

Your memory pipeline is accumulating noisy context instead of compounding knowledge. This happens when captured observations and session summaries lack proper promotion and resurfacing rules, causing repeated corrections to disappear before influencing future sessions.

The best way to classify captured knowledge is sorting it into compounding, dead, never captured, and noisy buckets. This diagnosis separates knowledge that actively guides future sessions from stale context that overwhelms useful signal.

Yes, you can apply memory decay rules to external memory stores. The audit traces how information moves from capture points including external stores into later sessions, designing resurfacing rules that prevent stale context from overwhelming useful signal.