Persistent memory for AI
Your AI keeps forgetting. Give it a memory that follows you everywhere.
One token gives every AI client — Claude Code, Codex, Cursor and more — the same living memory. Save it once on any machine, recall it everywhere in milliseconds, even offline.
- Cross-tool
- Cross-device
- Local & instant
- Works offline
Runs on your machine
Millisecond recall — no round-trip to the cloud.
The engine and your memories live right on your device. Store and recall in milliseconds, keep working with no network, and never wait on a server — the cloud is only for syncing.
Merged in the cloud
One complete memory, on every device.
Every device syncs up to the hub, which merges and curates your complete memory — then flows it back down, so recall on any machine returns everything you know, past and present.
One line to connect
Add xmemory as an MCP server. That is the whole install.
Once connected, just tell your assistant to set up xmemory — it runs memory_setup to install the local engine, migrate your existing notes, and turn on automatic recall.
$ claude mcp add --transport http xmemory \
https://xmemory.develop.cc/mcp \
--header "Authorization: Bearer <token>"
✓ xmemory connected — 5 tools ready.Works in the AI clients you already use
One central token connects them all — add xmemory as an MCP server and go.
Give your AI a memory that follows you.
One token, every device, one shared memory. Set it up in a couple of minutes.
Pay as you go · Curation is billed per item to your account — transparent usage, no subscription, no lock-in.