memory-hygiene

To reduce context costs, memory cleanup involves auditing and pruning stale entries from your memory file. This skill runs at session end or during cleanup to identify outdated references and duplicates, requiring explicit confirmation before deleting or archiving them.

Managing token growth requires memory budgeting, which analyzes your memory file to control size and optimize prompts. This skill implements size checks to detect duplicate entries and outdated references, preventing uncontrolled context expansion while preserving user preferences.

Memory hygiene does not automatically delete outdated memories without explicit user confirmation. It audits memory entries, detects duplicates and broken references, and requires your approval before proceeding with any deletion or archiving actions.

To validate external links and clean up broken references, this skill performs reference hygiene by checking stored memories against external resources. It prunes broken memories and identifies outdated references during a session audit.

You should run a memory audit at session start or session end to review and clean up memory files. Running it during compact or cleanup phases keeps context costs low by consolidating session learnings and archiving old entries.

You can preserve user preferences while pruning duplicate memory entries because this skill explicitly safeguards them during the cleanup process. It flags old learnings for archiving while retaining recent insights and user-specific settings.