Blog
Notes on running agents
What we learned building machines for coding agents, with the commands.
What an always-on agent machine costs
What an agent machine costs by the hour and month at shed's prices, what leaving it on adds up to, and how auto-sleep, stopping and deleting change it.
Move a Claude Code conversation to another machine
Where Claude Code stores each conversation, how to copy one to another machine and resume it with claude --resume, and what doesn't come along.
Keep Claude Code running after you close your laptop
Why a laptop agent stops when the lid closes or the Wi-Fi drops, and how to keep Claude Code working on a remote machine inside tmux.
Use Claude Code from your phone with Remote Control
How Claude Code's Remote Control lets you continue a session running on a remote machine from the Claude app, what it needs, and how shed sets it up.
Preview a dev server running on a remote machine
Open a dev server that runs on a remote machine through a private SSH tunnel or a public link, and get past the Vite host check that blocks tunnels.
Run your test suite on a bigger machine
Send a slow test suite from a laptop agent to a remote Linux machine with rsync, ssh and tmux, bring the results back, and work out the cost of each run.
Run several coding agents at once with git worktrees
Give each coding agent its own branch and folder with git worktrees, run each one in its own tmux session, and see where a laptop runs out of room.
Give your coding agent its own machine
What a coding agent can reach on your main computer, what Claude Code and Codex permission modes and sandboxes cover, and why a separate VM is cleaner.