What problem does it solve? Over time, an AI agent's memory files accumulate duplicates, stale facts, and outdated time references, making it hard for future sessions to orient quickly. This Skill performs a structured reflective pass that cleans up the memory directory so a new session can understand who the user is, what they work on, and how they like things done without re-asking. ## Core Features & Use Cases - Duplicate Merging: Identifies memory files describing the same person, project, or preference and combines them into a single richer file. - Staleness Cleanup: Separates durable knowledge (preferences, working style, key relationships) from dated items (finished projects, passed deadlines) and retires or folds the dated ones. - Time Reference Normalization: Converts relative phrases like "next week" or "by Friday" into absolute dates so memories stay readable later. - Index Maintenance: Keeps MEMORY.md under 200 lines and ~25KB with one-line entries linking to topic files. - Use Case: After weeks of sessions, your memory folder has 30 overlapping files. Run this Skill to merge them into a tight set of durable memories and a clean index. ## Quick Start Ask the agent to run a memory consolidation pass over its memory directory to merge duplicates, retire stale entries, and tidy the MEMORY.md index.
