How to Share MCP Server Configs Across Machines

MCP (Model Context Protocol) servers extend AI coding tools like Claude Desktop and Claude Code with custom capabilities. A filesystem MCP server gives Claude access to read and write files. A database server lets it query your PostgreSQL instance. A GitHub server lets it manage issues and pull requests. You can build custom servers for internal tools, documentation systems, or anything else you need Claude to interact with.

Each MCP server is defined in a configuration file with a command to start it, arguments, and often environment variables containing API keys or connection strings. Once you have five or ten MCP servers configured, manually recreating that setup on a new machine is tedious and error-prone. Worse, these configs contain secrets that should never be stored in plaintext or committed to version control.

Claude Desktop stores MCP server definitions in its main config file. On macOS, that is ~/Library/Application Support/Claude/claude_desktop_config.json. On Linux, it is ~/.config/Claude/claude_desktop_config.json. Claude Code stores its MCP settings within ~/.claude/.

Notice the API keys inline in the config. This is why syncing MCP configs requires encryption — these are not files you want sitting in plaintext anywhere.

ConfigSync has dedicated modules for both Claude Desktop and Claude Code that handle encryption automatically. When you add these modules, any config files containing secrets are encrypted before they leave your machine.

MCP configs often reference local filesystem paths. A filesystem server might point to /Users/jane/projects on your Mac but /home/jane/projects on your Linux workstation. ConfigSync template variables handle this automatically.

The {{home}} variable resolves to the correct home directory on each machine. On macOS it becomes /Users/jane, on Linux /home/jane. You can define custom variables too for paths that differ in more complex ways between machines.

One of the strongest use cases for syncing MCP configs is team standardization. When your team uses the same MCP servers — a shared documentation server, internal API server, or database query server — you want everyone to have the same setup.

The team lead maintains the canonical MCP server configuration. When a new server is added or an existing one is updated, they push the change and everyone pulls. API keys in the config are encrypted with each person's master password, so sharing the config does not mean sharing credentials. Each team member sets their own API keys after the initial pull if the keys are personal.

For shared team API keys, store them as ConfigSync secrets and reference them with template variables. This way the key is encrypted in the vault and injected into the config at pull time.

MCP server configs evolve as you add new servers, update API keys, and modify parameters. ConfigSync tracks these changes over time with version history, so you can see what changed and roll back if needed.

This is particularly valuable for MCP configs because a misconfigured server can break your Claude workflow entirely. If an update introduces a problem, rolling back to the last working configuration takes seconds instead of manual debugging.