[FEATURE] Claude Code memory and settings don't sync across devices — no supported path to enable this

Resolved 💬 2 comments Opened May 30, 2026 by PGarfunkel-Simpro Closed Jun 3, 2026

Preflight Checklist

  • [x] I have searched existing requests and this feature hasn't been requested yet
  • [x] This is a single feature request (not multiple features)

Problem Statement

Claude Code stores memory and global settings in ~/.claude/ (local user profile), which doesn't sync across devices by design. For users who work across multiple machines (e.g. home laptop + work laptop), this means:

  • Memory written on device A is invisible on device B
  • Global settings (.claude.json) and tool auth tokens (.clasprc.json) are siloed per device
  • Starting a session on a second device feels like starting from scratch

What we had to do to fix it:
We manually set up Windows directory junctions and file symlinks to redirect the local paths to a shared OneDrive folder:

  1. Replace ~/.claude/projects/<project>/memory/ with a junction pointing to OneDrive
  2. Replace ~/.claude.json with a symlink pointing to OneDrive
  3. Replace ~/.clasprc.json with a symlink pointing to OneDrive

This required admin PowerShell, Windows-specific commands, and troubleshooting across two machines with different usernames and shell environments. It took the better part of a session to get right. A non-technical user should not have to do this.

Who this affects:
Any Claude Code user working across more than one device — which is likely a significant portion of the user base.

Proposed Solution

Add a setting in Claude Code that lets users specify a custom sync path for memory and global config — pointing to OneDrive, Dropbox, iCloud, or any synced folder. One setting, no command line required.

Alternative Solutions

_No response_

Priority

Medium - Would be very helpful

Feature Category

Configuration and settings

Use Case Example

Working on projects across multiple devices (work and home laptop)

Additional Context

_No response_

View original on GitHub ↗

This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗