There’s a thing that happens in tech where a capability goes from “theoretical concern” to “completely normal” without anyone really deciding that transition was okay. I’ve been watching it happen with AI agents over the past year or so, and it’s starting to sit uncomfortably with me.

Not long ago, the conversation was about chatbots giving wrong answers. Hallucinating citations. Confidently explaining that the capital of Australia is Sydney. Embarrassing, but contained. The blast radius was: someone got bad information and maybe acted on it. Bad, but recoverable.

There are two reasons to pay attention to the Kavita v0.9.0 release. One is urgent. The other is genuinely good news.

Start with the urgent one. Two medium-severity CVEs have been disclosed against Kavita. The details are still making their way through the official CVE publication process, but the fix is already out. If you’re running Kavita on your home server or NAS, update to v0.9.0 now. Medium severity doesn’t mean ignore it; it means the window between disclosure and exploitation is shorter than you’d like. Self-hosted software is easy to neglect on updates precisely because it feels like it’s just sitting quietly on your network, not bothering anyone. That’s also the moment it becomes a problem.

There’s a particular kind of frustration that builds up slowly when you’re dealing with the modern web. You need to do something simple — resize a photo, strip some metadata, blur a face — and suddenly you’re being asked to sign up for a free trial, verify your email, and “unlock premium features” just to do what should take thirty seconds. It’s exhausting. And it’s gotten worse, not better.

So when I stumbled across a project this week — essentially Stirling-PDF but built for images — I found myself genuinely interested. The pitch is clean: one Docker container, browser-based, everything runs locally, your files never leave your machine. Thirty-plus tools covering the usual suspects like resize, crop, rotate, compress, and convert, but also some more interesting stuff like background removal, face and licence plate blurring, OCR, and object erasing. The developer is building it openly, asking for feedback, and has explicitly said they’re not interested in making it another “AI-wrapped gimmick or subscription trap.”

The recent Tesla recall of 700,000 vehicles for a tire pressure monitoring issue has sparked an interesting debate about what constitutes a “recall” in our increasingly software-driven world. While traditional recalls often meant bringing your vehicle to a dealership for hardware fixes, Tesla’s solution is a simple over-the-air software update that most owners will barely notice.

Working in software development, I find it fascinating how the automotive industry is grappling with this shift. The term “recall” carries heavy implications of faulty hardware and safety risks, yet here we have an issue that’s more akin to a smartphone app update. The specific problem - the tire pressure warning potentially not persisting after a system reboot - is certainly worth addressing, but hardly the kind of critical safety concern that traditionally prompted recalls.

Remember when having a home server meant running a simple file share and maybe a Plex server? Those days seem almost quaint now. The self-hosting community has evolved dramatically, and this week’s developments really highlight how far we’ve come.

The latest buzz around Glance, a multi-purpose dashboard and feed aggregator, caught my attention during my morning batch brew. What fascinates me isn’t just the tool itself, but how we’re now effectively creating dashboards to manage our dashboards. It’s like inception for home lab enthusiasts, and I’m here for it.

Standing in my home office, sipping my flat white and staring at the small cluster of Raspberry Pis humming away on my desk, I’ve been thinking about the state of self-hosted software. The topic caught my attention during a fascinating discussion about what tools people wish they could self-host but can’t find decent alternatives for.

What struck me most wasn’t the list of missing tools, but rather how far we’ve come. The open-source ecosystem has matured significantly over the past decade. Remember when running your own server meant endless nights of troubleshooting and a PhD-level understanding of Linux? Those days are largely behind us.