migrate-claude-memory
Inventory, clean up, and migrate Claude Code auto-memory stores found under ~/.claude/projects/<munged-path>/memory/ on this machine. Use this skill to list every memory store with its decoded project path, file count, size, and freshness; to identify ORPHANED stores whose original workspace no longer exists (so a human can review and delete them); and to migrate a chosen store into a repo's git-tracked .claude/memory/ directory so the memory travels with the repo across machines and is visible to hosted/cloud Claude sessions. Trigger on requests like "clean up claude memory", "migrate claude memory", "inventory memory stores", "orphaned memory", "sync memory across machines", "make memory portable", or any mention of ~/.claude/projects or autoMemoryDirectory. LOCAL-ONLY: this skill reads and writes files under this machine's ~/.claude directory and CANNOT run in a hosted/cloud Claude session that has no local ~/.claude on disk — do not invoke it there.
- Compatibility
- Requires bash, GNU coreutils (find, stat, du) and read/write access to ~/.claude/projects on the local machine; local execution only — memory stores are machine-local and this skill cannot run in a hosted/cloud session without that directory present.
Pinned to revision 400fbf01d7bb, so it is the text this page describes rather than whatever the author pushed since.
Files
- skills/migrate-claude-memory/SKILL.md
- skills/migrate-claude-memory/scripts/memory-inventory.sh
- skills/migrate-claude-memory/scripts/memory-migrate.sh
Every link opens the file at its source, pinned to the revision this page describes.