Sno Memory for Cursor

Gives the Cursor IDE and the Cursor CLI the same encrypted, local memory that Claude Code and Codex use through Sno: a conversation starts with what the repository already established, relevant memory arrives with each message, and what the conversation settles is stored for the next one, whichever agent opens it.

New in 2.0: Tag Team. Your Cursor status line now says when any agent on your squad needs you, and sno squad shows the whole picture. What changed.

Get started

1. Follow the memory service setup.

2. Run sno setup --harness cursor. It writes Cursor hook entries into ~/.cursor/hooks.json beside any entries other tools wrote there, allows sno in the Cursor CLI, and installs the sno-mem-cursor skill.

3. Open a Cursor conversation inside a git repository and try:

   sno memory doctor --harness cursor
   sno memory remember --harness cursor "Prefer tabs in this repository"
   sno memory recall --harness cursor "indentation"

A conversation outside a git repository reads and writes nothing. A conversation with several folders open belongs to the repository of the first folder.

Walkthrough: Cursor onboarding.

Under the hood

This package is the Cursor side only: Cursor hooks and the memory commands. The memory itself lives in @snoai/memory, a local service on 127.0.0.1 that owns the encrypted store.

License

Licensed under Apache-2.0.