Optimizes Claude's contextual memory by identifying and reviewing low-confidence or outdated stored information.
Memory Context Cleanup is a specialized utility for Claude Code designed to maintain a high-quality knowledge base by pruning the AI's stored memories. It provides an interactive workflow to review information that Claude has flagged with low confidence, allowing you to archive obsolete data, reinforce important patterns, or skip ambiguous entries. This process ensures that Claude's responses remain accurate and relevant by preventing stale preferences or incorrect context from cluttering the active memory during development tasks.
Key Features
01Quick archiving of stale or incorrect preferences
02Interactive review loop for low-confidence memories
030 GitHub stars
04Customizable confidence threshold filtering via CLI options
05Direct reinforcement to boost confidence in valid data
06Detailed summary reports of memory state changes
Use Cases
01Cleaning up misinterpreted context from previous chat sessions
02Pruning outdated coding preferences after switching project styles
03Optimizing AI accuracy by reinforcing valid but low-confidence memories
