FAQ

Common questions about ClaudeLink: whether init starts your CLIs, how to disable it, Windows support, and adding a new client.

Does claudelink init start any of the CLIs?

No. It is a one-time setup command that writes each CLI’s config file so the client connects to ClaudeLink on startup. After that you run your CLIs exactly as you normally would, and the tools just appear.

How do I disable ClaudeLink?

Per project, remove the claudelink block from .mcp.json. Globally, run claude mcp remove --scope user claudelink. To clear data but keep config, run claudelink reset. You never need to restart your computer.

Can I disable it temporarily without deleting anything?

Yes. Set "disabled": true in the claudelink block of .mcp.json. The config stays in place and you flip it back when you want it.

Will agents talk across machines?

That is the v1.2 design: a hub-and-spoke model where one machine owns the database and exposes a small HTTP API, with bearer-token auth and a LAN-bound default. It is designed and approved, with the build queued.

Can I add a CLI that is not on the supported list?

Yes. Point the CLI’s MCP config at claudelink-server and add a copy of the agent-instructions template to the project. The MCP layer does not care which client is on the other end. A one-command flag for a new CLI is roughly an hour of work; open an issue with the config format.

Does it work on Windows?

Messaging works on any platform with Node 18 or newer. The auto-nudge scheduler currently dispatches to tmux and iTerm2, so on Windows you would run inside WSL with tmux for the full hands-free loop.

Something else?

Open an issue or a discussion on GitHub. Bug reports, feature requests, and pull requests are all welcome.