Skip to main content
RemoteMux lets you run AI coding agents on a remote VM that keeps working when your laptop sleeps, your Wi-Fi drops, or your battery dies.
This guide assumes you have already run rmux env new and rmux workspace new main. If not, start with Quick Start.

Claude Code

Attach to your workspace and start Claude:
Give Claude a task, then detach with Ctrl+B, D. The VM keeps running. Come back later:

Claude Code Auth

RemoteMux performs a best-effort Claude Code auth sync before rmux env new, rmux workspace attach, rmux workspace new, and rmux workspace run. If Claude is authenticated locally, the remote workspace can usually use it immediately.

Codex

Detach and reconnect the same way. Any CLI agent that runs in a terminal works.

Codex Auth

RemoteMux also performs a best-effort Codex auth sync before rmux env new, rmux workspace attach, rmux workspace new, and rmux workspace run. When Codex is configured locally, RemoteMux syncs the first available source in this order:
  1. CODEX_API_KEY
  2. OPENAI_API_KEY
  3. ~/.codex/auth.json (including ChatGPT OAuth sessions)

Parallel Agents

Run multiple agents on the same project simultaneously, each in its own isolated workspace:
Start a separate agent in each:
Each workspace gets its own git worktree, so multiple agents can edit the same repo without merge conflicts.

Monitor progress

Move changes back

After agents finish, use git to review, push, and merge their branches from the main workspace or from your local checkout.

Secrets

Agents often need API keys. Set them once per environment:
Secrets are available as environment variables in all workspaces.

Tips

  • Use named workspaces for agents — keep main clean for manual work.
  • Detach, do not killCtrl+B, D detaches without stopping anything. exit or Ctrl+D ends the session.
  • Expose ports — if an agent starts a dev server, use rmux env expose 3000 to get a public URL.
  • Stop compute when donermux env down stops billing while preserving workspace state.
Last modified on April 9, 2026