What problem does it solve? When a memory system routes user prompts to stored memory files, there is no built-in way to know whether the routing is accurate, whether relevant context is being missed, or whether retrieval is wasting tokens. This Skill audits the multiplai-context plugin's retrieval logs and turns them into measurable quality metrics. ## Core Features & Use Cases - Quantitative log analysis: Computes volume, routing rates, file distribution, token cost, dedup effectiveness, pre-filter rates, and error rates across daily context-router log files. - Qualitative sampling and false-negative detection: Grades sampled routing decisions as correct, over-broad, or wrong, and checks NONE-routed entries with personal keywords for missed retrievals. - Watermark-based delta tracking: Partitions pre/post-fix data using a watermark file so each report compares current metrics against the previous baseline. - Use Case: After deploying a routing fix to the context-router hook, run this Skill to verify that false negatives dropped and that retrieval size stayed under the 15K byte cap. ## Quick Start Ask the assistant to analyze memory retrieval quality and report how the memory loader has been performing since the last analysis.
