Your server runs eight things.
Know when one dies.
Fusebox connects to your server over SSH and works out what it's actually running. It checks that each service really answers — not just that the process is up — and tells you when one goes down.
Get it on the App StoreFree for 7 days, everything unlocked. Then $7.99 once. No subscription.
What it looks like
It finds them by itself
No configuration. Connect, and the services you're running show up with their own cards.
"Running" and "working" are different things
systemctl says active and docker ps says Up while the service behind them stopped answering hours ago. That's the failure that actually bites, and process state can't see it.
Fusebox asks the service itself, on your server's own loopback, and reports one of four states: responding, not responding, running but unverified, or stopped — with a timestamp, so a check from an hour ago doesn't read as current.
It doesn't cry wolf
A container you stopped on purpose is not a problem, and Fusebox won't treat it like one. It reads the intent your system already declares — whether a unit is enabled at boot, whether a container has a restart policy — instead of guessing.
Something declared always-on that isn't running gets an alert. Something you shut down yourself stays quiet.
When something breaks
- Restart the container or the unit from the same screen
- Read
journalctland container logs without opening a terminal - A Health Score that inspects the server, grades it, and gives you the exact command for each finding
- A notification when a background check finds a service down — local, fired by the app itself, since there's no server to push from (see the limits for how often iOS allows that)
How it connects
Plain SSH. Nothing installed on the server — no agent, no daemon, no cron job.
No Fusebox server. Your phone talks to your server and to nobody else. No account, no telemetry.
The host key is verified before any credential is sent. Pinned on first connection, like known_hosts. If it changes, the connection stops and Fusebox explains why instead of handing your password to whoever answered.
Credentials live in the iOS Keychain.
What it doesn't do
iPhone only. No iPad layout yet, no Mac app, no Android.
Background checks are not real time. With the app closed, iOS decides when it may run — typically every 15 to 60 minutes, sometimes not at all if you rarely open the app or Background App Refresh is off. Anyone promising continuous monitoring from an iOS app is describing something the platform doesn't allow.
Same network, or a VPN. Fusebox speaks straight to your server with nothing in between, so from outside its network you need something like Tailscale or WireGuard. The app explains how when it detects the situation.
Which means alerts don't follow you everywhere. Off that network with the app closed, a background check can only reach your server if your VPN is always-on — Tailscale can be. Without that, the check can't connect and you won't get the alert. It's the honest consequence of having no server in the middle.
RSA keys only, or a password. The SSH library underneath doesn't parse Ed25519 or ECDSA yet. If your server only accepts an Ed25519 key, Fusebox can't connect to it today.
It's closed source. The commands page exists because that's a fair thing to want to check.
Questions people actually ask
Why not just use Uptime Kuma?
If you already run Kuma on a second machine, keep it — nothing on a phone beats a box that checks every 60 seconds. Fusebox is for the server that has nothing watching it: no setup, it works out what you're running by itself, and when something is down the restart button, the logs and a terminal are on the same screen as the alert. Kuma tells you it's down; it can't bring it back.
Does it have to be a Raspberry Pi?
No. Fusebox runs on any Linux host over SSH — a VPS, a NAS, a Proxmox container, an old laptop, or a Pi. It detects what it's connected to and adapts: temperature, throttling and undervoltage checks only show up when the host actually is a Raspberry Pi.
Multiple machines?
Yes — included in the one-time unlock, along with the container and service controls, alerts, history, terminal and saved commands. After the trial ends without paying, the app keeps working as a single-machine monitor.
Why is it paid, and why one payment?
There's no server, so there's nothing a subscription would be paying for. Seven days free with everything unlocked, then $7.99 once.
Why isn't it on the App Store in Europe yet?
EU distribution comes with requirements for individual developers that I haven't been able to meet yet. If you're in the EU and want Fusebox, write to me at gortega+fusebox@gmail.com or on LinkedIn and I'll tell you the moment that changes.