Claude Code memory synchronization guide for multi-machine setups

A technical guide details methods for synchronizing memory across multiple machines when using the Claude Code command-line interface (CLI). Because Claude Code stores project-specific memory and global preferences locally rather than in the cloud, users working across different hardware, such as a MacBook Pro and a Mac mini, face challenges with inconsistent context.

To resolve this, the proposed solution involves creating a private Git repository to manage specific configuration files and directories. Key components for synchronization include project memory files, custom skills, global preferences (CLAUDE.md), global settings (settings.json), and MCP configurations. The guide warns against synchronizing entire directories like ~/.claude, as they contain large session histories, local runtime states, and potentially sensitive plain-text access tokens that should not be shared or pushed to repositories.

For directories like 'skills' and project memory, the method utilizes symbolic links to ensure real-time updates within the Git repository. For individual files that the CLI might overwrite, a manual copy-and-sync approach based on modification timestamps is recommended to prevent broken links.