Urgent.News

What's breaking now, across thousands of outlets.

AI

I built a self-hosted deploy tool that looks like tmux (and can be driven by an AI agent)

I deploy a handful of projects to a handful of my own servers. Not enough to justify Kubernetes, too many to keep doing ssh , git pull , docker compose up -d and hoping. I didn't want a PaaS, a cloud control plane, or an agent running on every server. So I built deploy.me : a local, Dockerized admin app. You run it on your own laptop (macOS, Linux, or Windows with WSL2) at…

The author built a self-hosted deploy tool called deploy.me, which looks similar to tmux and can be controlled by an AI agent. The tool is local, Dockerized, and runs on a user's laptop (macOS, Linux, or Windows with WSL2). It retrieves server information from ~/.ssh/config and imports projects from GitHub or a local folder. Deployments occur using only ssh, rsync, and docker compose on the other end.

The tool is MIT licensed and allows for blue/green deployments with minimal downtime. The user interface is a tmux session with panes, a status bar, and a shell pane for real work. An AI agent named Claude can be integrated to explain deployment failures and plan deployments, although by default, the agent operates read-only. Security measures include not storing secrets in arguments or logs, requiring confirmation for sensitive commands, and trusting host keys.

The tool's codebase was primarily developed using Claude Code, with a focus on fixing suspected bugs through testing. The user can access deploy.me through a one-time link after setting up a passkey or password. The project's repository is available at https://github.com/KielD-01/deploy.me, and additional documentation, including screenshots and an MCP guide, can be found on GitHub.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

Shipping claude-shiploop

Every case study on this site runs through the same loop: classify the request, write the spec, build to the spec, measure the result, promote what repeats, forget what doesn't.

More from Saturday 17 October →