brewdoc:memory

Reduce token usage by removing duplicate memory entries, migrating persistent information to CLAUDE.md files, and compressing remaining context to optimize Claude Code memory. This process eliminates token bloat and improves AI response efficiency across sessions.

The best way to migrate persistent rules is to move architectural decisions and fixed configurations out of memory entries and into dedicated CLAUDE.md files and rule configurations. This ensures only essential, session-specific context remains active in memory.

Yes, optimizing Claude memory requires file system access for reading and writing configuration and memory files. The process validates memory structures, compresses entries, and migrates persistent information to CLAUDE.md files on disk.

Validate memory files by running an optimization process that checks the final structure for broken references and ensures memory files are well-formed. This validation step occurs after duplicates are removed and remaining entries are compressed into token-efficient formats.

You should regularly clean up Claude Code context when memory entries accumulate duplicates or persistent rules that belong in CLAUDE.md files. Regular optimization prevents token bloat, reduces costs, and improves response times by ensuring only session-specific context remains.