Your hardware.
Our control plane.

Logs stream straight from your server while you watch. CPU, memory, restarts and requests are kept by the hour for seven days.

Every ten minutes Eris requests each of your app's names from both coasts of North America, both halves of Europe, Asia-Pacific and Oceania. The Dashboard shows each answer, how long it took and which city it came from.

Any HTTP answer counts, a 500 included: Eris checks that the route to your app works, not your code. The requests carry User-Agent: Eris-Reach/1, so you can find them in your own logs.

Eris runs the database as a container on the same server, on a network of its own. It lives on your disk, and one database can serve several apps.

Attach an app, and its next deploy finds DATABASE_URL or REDIS_URL in the environment. You pick the version once, and Eris never changes it.

Name an S3-compatible bucket you own, and Eris dumps Postgres, MySQL or MariaDB into it once a day. The dump goes from your server straight to the bucket.

Give a task a name, a schedule and a command, like php artisan schedule:run every minute. Each run starts a fresh container from the image and variables your app runs with, on its networks, so it reaches your database and your volume.

Pick a schedule from a list or type five cron fields, in UTC. The agent on your server fires each run on its own clock, so tasks keep running while Eris is down. A run is stopped after an hour.

Run a task now, pause it, or read its output. Eris records how each of the last 50 runs ended. Output is read from your server while you look, and only the last run's is kept.

Activity lists deploys, failed task runs, backups, outages and a filling disk, across every app and server, for 30 days. Each row says what set it off, whether you, a push, a schedule, Eris or Claude Code, and opens the screen where it happened.

Filter to one app or to problems only. A deploy still running turns into its result without a reload.

Eris checks every minute. When a server, an app or a deploy needs you, it sends one email. When the problem clears, it sends nothing: the Dashboard already shows that.

Turn any of them off in the Dashboard, or stop all email with one click from any message.

Turn the firewall on, and the server accepts only what your rules open, plus 80 and 443. Each change runs on trial: if Eris loses the server or port 80 stops answering, the agent puts the previous rules back.

Lock yourself out of SSH, and the Dashboard is still your way back in. Once the firewall holds, Eris turns off the ufw or firewalld it replaced.

Eris speaks MCP. Connect Claude Code or claude.ai, and it reads your servers, your apps, their last ten deploys and a deploy's log.

It only reads: no tool deploys, changes or deletes. You allow each assistant in the browser and revoke it in the Dashboard. Eris disconnects one that goes unused for 30 days.

Each one lives in the Dashboard, on the app or the server it belongs to.

Coolify and Dokploy install theirs on your server: you update it, you fix it, and when the server dies, it dies too. With Eris, your server runs an agent and your containers. If Eris goes down, your apps keep answering.

Wiring a VPS by hand, the Dockerfile and the nginx config take an afternoon. Certificates that renew and deploys that fail safely take years. Here, a failed deploy leaves the running app untouched.

Your app runs on your hardware, and its traffic never passes through Eris. Its image is built on Eris's builders and stored in Eris's registry.