memory-optimize

To reduce token count in Claude Code memory files, you can use an interactive optimization process that identifies redundant data, migrates misplaced rules to persistent config files, and compresses content into efficient formats like tables. This can lower token usage by 30-50%.

Your Claude memory file slows down context loading and increases API costs due to growing inefficiency from redundant entries, misplaced rules, and inefficient formatting within unoptimized auto-memory files, which require more tokens to process on each request.

To remove duplicate rules from auto-memory files and CLAUDE.md, an optimization process analyzes your project and global rules to identify overlapping entries, removes the redundancies from memory files, and intelligently migrates the rules to persistent configuration directories.

The best way to compress Claude memory file content for token efficiency is reformatting remaining entries into denser structures like tables after analyzing and migrating misplaced rules, which maximizes token reduction while preserving the required context.

Yes, you can migrate architectural decisions from auto-memory files to persistent Claude rules. The optimization process intelligently moves relevant rules from auto-memory into CLAUDE.md, .claude/rules/, or ~/.claude/rules/ to maintain persistent configurations.

To validate and cleanup orphaned files after optimizing Claude memory, the process checks for broken references and orphaned files in the final validation step to ensure the remaining memory structure is clean and fully functional.