Claude Code configures a home server and backups
Jordan Fulghum installed Claude Code on an Ubuntu mini PC and used it to configure containerized personal services, a monitoring dashboard and local and S3 backups.
Original prompt
Shared by the source author. See source notes for context.
Create a Docker Compose file
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Jordan Fulghum
- Original account
- Read the source ↗
- Published
- 5 Jan 2026
- Source link checked
- 15 Oct 2026
- Reported result
- Fulghum reports fourteen running containers, a custom monitoring dashboard, daily USB backups and weekly S3 backups.
- Human involvement
- Fulghum installed the base system, provided access, attached storage and created the AWS IAM user; backup restoration is not demonstrated.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Install Linux, Tailscale and Claude Code on the home server.
- Describe required services, storage persistence, reverse proxy and restart behavior.
- Ask Claude to build a dashboard showing service state and backup logs.
Jordan Fulghum wanted to run personal services at home without spending every evening assembling configuration files. He installed Ubuntu Server, Tailscale and Claude Code on a Beelink mini PC, then described the services and behavior he wanted from the machine.
Claude Code worked directly on the server. Fulghum asked it to set up Docker, Compose files, persistent storage, Caddy routing and restarts after outages. His account lists working services including Vaultwarden, Immich, Plex, Readeck and Uptime Kuma. He also asked for a single monitoring page; Claude built a local Go and Svelte application with service links, status information and backup monitoring.
The backup work gives the account a concrete sequence. Fulghum attached a USB drive and asked for daily local backups. He then created an AWS IAM user himself, installed the AWS CLI and requested weekly S3 backups. Claude suggested Glacier storage and configured remote backups with dashboard logs.
Fulghum still supplied the hardware, operating system, access and cloud credentials, and he describes returning through SSH when something breaks. The account reports working services and scheduled backups, but does not document a restore exercise or independent security audit. Its useful lesson is to give a terminal agent the desired service behavior and inspect the resulting operations, including persistence and recovery.
Prompt
Original prompt
Source: Blog · Jordan Fulghum
Frequently asked questions
Where did Claude Code run?
It was installed directly on the Ubuntu home server, which Fulghum accessed over his private Tailscale network.
What did the owner do for the remote backup?
He created an AWS IAM user and installed the AWS CLI before asking Claude to configure weekly S3 backups.
Does the source demonstrate a successful restore?
No. It reports configured local and remote backups with monitoring and logs, but does not describe a restore test.
More details
Remaining steps
- Attach a backup drive and create AWS access before requesting local and remote schedules.
About this account
Claude Code translated service and backup goals into server configuration and built a monitoring application on the machine.
Firsthand source opened over HTTP and read on 2026-10-15. Publication date comes from the source page. Outcomes are author-reported and were not independently reproduced. Services and backups are owner-reported; no restore test or security audit documented. Original prompt box contains a short excerpt of the author's published instructions.
Credit: Jordan Fulghum
Source published 5 Jan 2026
Source checked 15 Oct 2026
Tools and topics
Claude Code · Ubuntu Server · Docker Compose · Tailscale · Caddy · AWS CLI · Amazon S3
Home server · Personal services · Backups
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.