Cloud in a Bottle wants self-hosting to feel like an app store, not a sysadmin job

A new open-source project called Cloud in a Bottle wants to make running your own personal cloud as easy as using a smartphone. It launched on 5 September with a pitch that will sound familiar to anyone who’s tried self-hosting before: existing tools are either abandoned, insecure, or force every app to live in its own disconnected silo. Whether it actually delivers on that is, for now, a claim rather than a proven fact.

What Cloud in a Bottle actually is

Per the launch post, the setup is fairly simple in concept: an Ubuntu machine running a web server that hosts a dashboard and routes traffic to containerised apps. Founder Zack Polizzi argues that today’s self-hosting tools all fall short in their own way - Sandstorm.io was abandoned years ago, Nextcloud has drifted towards enterprise and become sluggish, YunoHost runs apps unsandboxed on the bare host (so one dodgy app can compromise everything), and Coolify treats every hosted app as an island with its own separate login.

Cloud in a Bottle’s answer is single sign-on across all your self-hosted apps, plus an opt-in framework letting apps share data and permissions with each other - the sort of thing Android or iOS offers its apps via shared system APIs, but for software you run yourself on your own box. The apps run in “rootless, hardened” containers, which the project says means existing software can run with minimal changes while staying reasonably sandboxed.

The bigger argument behind it

The pitch isn’t just technical - it’s about incentives. Polizzi’s broader argument is that cloud software became centralised because individuals never had an easy way to host their own instance of anything, leaving the bill (and the incentive to monetise via ads and data harvesting) with whoever runs the one central version everyone uses. Self-hosting, in theory, puts you back in charge of your own instance and your own data. It’s a compelling narrative, but it’s also the same one self-hosting advocates have been making for over a decade, including the very projects Cloud in a Bottle argues have failed to pull it off.

So what actually happened at launch, and does it undercut the pitch

The project picked up 303 points and 130 comments on Hacker News - solid traction for a brand-new tool. But a chunk of that discussion wasn’t about the software at all. A commenter flagged that an account called “adl-collab” had been quietly filing GitHub issues on other projects’ repos requesting Cloud in a Bottle compatibility, with no disclosure of any connection to the project. Digging further, commenters linked the account to Andrew Laack, described as the project’s most active recent contributor.

Polizzi, posting as the project’s author, confirmed the issues likely came from his team and apologised, calling it “a mistake, not some huge spamming campaign” and putting the number at 11 issues total. He also noted such outreach was premature for a tool that doesn’t yet have a meaningful user base. It’s a minor scandal as these things go, but on a project pitched around trust, transparency and giving users back control, an unmarked promotional push on day one is a rougher start than the software itself deserved.

What this means for you

Nothing here is proven at scale yet - no independent security review, no track record, no evidence the “hardened” containers hold up under real attack. If you’re the sort of person who already self-hosts, it’s worth a look and a healthy dose of patience. If you’re not, there’s no reason to touch it yet: wait and see whether the promises about security and usability survive contact with actual users, rather than just a launch post and a lively comments section.