memory-hygiene

The memory-hygiene skill is designed to maintain the cleanliness, consistency, and efficiency of memory storage within an AI system. It addresses common issues such as memory bloat, duplicate entries, stale data, and schema inconsistencies, ensuring that memory remains queryable, compact, and reliable. By automating critical hygiene operations, this skill prevents performance degradation and supports high-quality data management, reducing the risk of errors or inefficiencies in memory-dependent processes.

The skill provides several key capabilities to uphold memory integrity. It can compact memory files by deduplicating identical or highly similar entries, bucket entries by topic for faster retrieval, and generate detailed compression reports. It validates memory entries against a defined schema to ensure proper structure, data types, and confidence scores. The skill also expires stale entries based on TTL values, archiving them if needed, and supports merging duplicates while preserving metadata and retaining the highest confidence versions. Safety measures include automated backups, logging, dry-run previews, atomic operations with rollback, and full provenance tracking.

Memory-hygiene is particularly useful for AI system developers, data engineers, and memory-intensive applications that require reliable and maintainable memory storage. It is ideal before bulk memory operations, during periodic maintenance, or after detecting inconsistencies and duplicates. Use cases include managing large datasets, ensuring data quality in long-running AI systems, performing schema validation, and automating memory housekeeping tasks to reduce manual intervention and maintain system performance over time.