Do I need to actively manage it all the time?
No. The whole point is that codemem captures useful context automatically while you work. You install it, wire it into your agent workflow, and mostly leave it alone until you want to recover something later.
Do I need sync to use it?
No. codemem is useful as a local-first tool on one machine. Sync is there when you want continuity across devices.
What is special about the sync story?
codemem stays useful on one machine, then extends that memory across your own machines with peer-to-peer sync. You do not have to choose between local-only and shipping everything into a hosted service. It stays private by default, and you can choose what gets shared.
Is this only for AI coding agents?
Right now, that is the main focus. codemem is designed for OpenCode, Claude Code, and adjacent AI coding-agent workflows where useful context disappears between sessions.