What problem does it solve? Long-term AI memory stores accumulate duplicates, stale facts, leaked credentials, and vague entries over time, degrading retrieval quality. This Skill runs a structured four-phase consolidation pass that cleans, merges, and rewrites memories into durable, well-categorized knowledge. ## Core Features & Use Cases - Duplicate Merging: Detects memories expressing the same fact and merges them atomically via memory_update, preserving edit history and temporal anchors. - Credential and Noise Removal: Deletes API keys, tokens, raw tool output, heartbeat records, and TTL-expired operational or project memories. - Quality Rewriting: Converts first-person entries to third person, adds temporal anchors, fixes miscategorized memories, and compresses entries to 15-50 words. - Use Case: After weeks of an agent storing observations, invoke this Skill to review all memories, merge five overlapping notes about a user's Python preference into one clean entry, and purge expired session data. ## Quick Start Ask the agent to run a memory consolidation pass to clean up, merge, and review all stored memories.
