memory-sync

To persist session learnings to CLAUDE.md automatically, use a memory synchronization skill that writes your debugging insights and milestone outcomes to persistent memory files in a forked context.

Auto-memory drift happens because the built-in auto-memory system cannot update CLAUDE.md, skill files, or project-local MEMORY.md directly, requiring a dedicated sync process to capture evolving session context.

You should sync session learnings to project memory files after reaching project milestones, completing sprints, or resolving notable debugging sessions to ensure persistent context stays current.

You can update CLAUDE.md and MEMORY.md without bloating the main conversation by running a memory sync operation in a forked context, isolating the file writing process from your active session.

The best way to manage memory persistence across project transitions is to trigger an auto-sync process that explicitly writes session learnings to CLAUDE.md and project-local MEMORY.md files.

The built-in auto-memory system does not update project memory files like CLAUDE.md or MEMORY.md, necessitating a separate memory-sync skill to define ownership and persist these specific learnings.