Actually more than saving memories that may be polluted, I go spec-driven: each new feature has its spec that serves both as a plan and doc reference plus for reusable stuff i create dedicated skills for that.
it's all plain markdown in ~/.claude/projects/<project>/memory/ so a private git repo works, you can also ask claude itself to backup your local claude code memory at the end of each day but may I ask what is the use case?
I vibe-coded my own tool to keep all my session/memories/... in sync across my laptops and push it to my NAS. Not released anywhere.
— https://dev-null.sh/@jubianchi/commits/01a179ad-a346-7186-8864-4d057b96f028
/dev/nullJulien Bianchi (@jubianchi) on /dev/null: “What's the projects/tools did you vibe-coded to help you in your daily…”What's the projects/tools did you vibe-coded to help you in your daily work? On my side, I built: * homesync: a tool to keep my two laptops in sync (dotfiles, clones repos, agents sessions, ...) *…