One command in.
Zero-downtime out.
Enroll a server with a single line. From then on every deploy is built off-box, swapped blue-green, health-gated and switched atomically — while the agent dials home outbound-only and your data never leaves the box.
Four moves. Nothing native to break.
Apps and backing services alike run as pinned containers. Here is exactly what happens between git push and live traffic.
Build off-box
Dockerfile-first, or auto-detected if you don't have one. The image is built on Sproobo's builders — your production server never thrashes compiling, and never runs a half-built state.
Blue-green swap
The new container starts alongside the running one. Both are healthy before anything shifts, so there is no window where the app is down by construction.
Health gate
The release only proceeds if it passes your health checks. Fail the gate and nothing ships — you're left on the last-known-good container, automatically, untouched.
Atomic proxy switch
Caddy flips traffic to the new container in a single move, with HTTPS and certificate renewal on ACME autopilot. Rollback stays armed if you need to reverse.
Reliability you can name.
Not adjectives — a pipeline. Every stage is observable and every stage can safely abort.
A single static binary that only ever dials out.
The Sproobo agent is one Go binary with no runtime dependencies. It opens a persistent outbound WSS/443 connection to the control plane — so there is nothing to expose, no inbound firewall hole, and no SSH port for the platform to sit on.
- outboundWSS/443 only — behaves like any HTTPS client.
- staticNo interpreter, no package tree, no drift.
- break-glassYour own SSH is never severed. Lock-out is impossible.
Control without the caretaking.
Docker everything
Apps and services run as pinned, choosable containers. Nothing is coupled to the host OS, so nothing native can break underneath you.
Data on your box
App data lives on bind-mounted host paths you can open, back up and reason about. Survives volume prune.
Zero-downtime deploys
Blue-green swap, health gate, atomic switch and armed rollback on every release. No maintenance windows.
Outbound-only agent
One static Go binary over WSS/443. No inbound ports, nothing to expose to the internet.
AI-native, safely
Drive infra via AI/MCP through the same audited API a human uses. Destructive ops are plan-first and approval-gated.
Migrate incrementally
Bring servers one at a time from Forge, Ploi or a hand-rolled VPS. No big-bang cutover, no all-or-nothing bet.
Domains & TLS on autopilot
Apex + www on every cert, automatic HTTP→HTTPS, shared wildcard certs for subdomain apps — and DNS-01 issuance for servers that aren’t reachable from the internet at all.
When a deploy fails, you see why
The failing container’s log tail is captured and pinned to the deployment itself. No SSH archaeology at 2am — the reason is right there next to the red X.
Static sites too
Point Sproobo at a plain HTML repo and it’s served as-is — no Dockerfile, no build step, dotfiles hidden by default. The same certs, domains and deploys as any app.