I've been using Claude Code across multiple repos and machines, and managing the config got messy fast—scattered .mcp.json files, duplicated slash commands, credentials in random places.
So I built claude-config: a simple framework to centralize and version control your Claude Code setup, the same way we do traditional dotfiles.
One bootstrap.sh, one config file. It symlinks commands, sets up MCP servers per-repo, handles credentials, and keeps everything in sync across machines.
GitHub: https://github.com/sumchattering/claude-config
Would love feedback, especially if you're managing Claude Code across multiple projects.
Cheers Sumeru
