Sproobo v6 is a from-scratch rebuild: the control plane, the on-server agent, the way apps and the databases behind them get provisioned, and the way we bill. It's running now. It runs alongside the previous system while servers move over one at a time, and it comes with something new: a Free plan that runs the whole product on one real server for €0. This post is the map. Each capability below links to the post that takes it apart in full.
Why we rebuilt
The short version, because there's a whole post about the long one. Every recurring outage in the previous Sproobo traced back to the same root cause: production behavior tied to a host's mutable, shared state. A package source fighting the distro's own. A PPA lagging a new OS release. A PATH resolving the wrong binary. Some other container seizing :80 before our proxy did. None of those were bugs, exactly; they were the cost of installing dependencies straight onto a live production host. v6 deletes that entire failure class by not putting dependencies on the host at all. The full postmortem is in Why we rebuilt Sproobo on Docker-everything.
What's new
Docker for everything. Every workload runs as a container with a version you pin yourself: your app, and every backing service behind it (PostgreSQL, MySQL, MariaDB, Redis, MongoDB). Nothing gets compiled on your box; images are built off-box and the agent only ever pulls a finished, tagged image. Durable data lives on bind-mounted host paths under /var/lib/sproobo, so it survives docker compose down -v and stays inspectable with plain ls and tar. See Docker-everything.
An agent that only dials out. The on-server agent never listens for anything. It dials home over WSS/443 and HTTPS, carries a rotating mTLS identity, and needs no inbound firewall hole. It's also structurally barred from ever severing its own channel or closing break-glass :22, so no operation can lock you out of your own server. See An agent that only dials out.
Deploys that explain their own failures. Releases are blue-green: a new container comes up and passes a health gate before the proxy upstream switches atomically, so a bad release never takes the running one offline, and rollback is instant. When a deploy does fail, the failing container's own log tail is captured and attached to the deployment record, no SSH required to find out why. See Deploys that explain their own failures.
Your infrastructure, one conversation away. claude.ai and ChatGPT connectors, a CLI with browser login, and an Agent Skill for coding tools all reach your infrastructure through the same audited, role-gated API a human uses. Reads are free; writes are plan-first; the one destructive operation never auto-applies. See Your infrastructure, one conversation away.
TLS for every topology. A public box, a VPN-only server, a fleet of subdomains, and an app behind someone else's load balancer each need a different certificate story. v6 ships all four as a per-app setting: HTTP-01, delegated DNS-01, wildcard, and external termination. See TLS for every topology.
What every plan guarantees
Some things aren't a tier. On every plan, from Free to Enterprise: we read your telemetry live and store none of it; break-glass SSH is never severed; your durable data lives on bind-mounted paths you own; and leaving is copying files off your own machine. Ownership isn't the upsell here, it's the floor.
Start free
The Free plan is €0 and runs the actual product, not a demo: one real server, three apps, two databases or services, two team seats, and one release kept for rollback. It's enough to enroll a box you already have and put something real into production. When you outgrow it, Pro is €19/mo (three servers, unlimited apps) and Team is €29/mo (unlimited fleet, the whole team); Enterprise adds a security review, a DPA, and an SLA.
Free €0 1 server · 3 apps · 2 services · 2 seats Pro €19/mo 3 servers · unlimited apps · 10 services · 5 seats Team €29/mo unlimited fleet · unlimited seats Enterprise custom everything unlimited · security review · DPA · SLA
Caps are per-organization, and billing runs through a Merchant of Record so VAT and invoices are handled for you. Downgrading never deletes anything: resources over a lower cap keep running, you just can't create new ones until you're back under it. The full table is on Pricing.
What the move means if you're on the old Sproobo
None of this shipped as a big-bang cutover. v6 runs alongside the previous system today, and servers move over one at a time by re-enrolling, so the new architecture proves itself against your real traffic before it has to carry all of it. Nothing changes under you until a server is re-enrolled; when it is, that box gets the whole v6 story above. Getting a first server on is the Getting started guide.