Everything that can be accomplished with traditional Infrastructure as Code (IaC) tooling can be done using System Initiative. This reference highlights key differences in methodology, workflow, and general capabilities. It also provides guidance on how to accomplish similar tasks in System Initiative to those you're familiar with in Infrastructure as Code solutions.

System Initiative isn't a DevOps tool. It's a new foundation for automation. It's not about replacing humans, or replacing your stack. It's about finally giving teams a way to make infrastructure work easier, safer, and collaborative.

I gave it a quick spin with System Initiative[1]. The combination solved a 503 error in our infrastructure in 15 minutes that took over 2 hours to debug manually.

It's pretty good! I wrote about a few other use cases on my blog[2]

1: https://systeminit.com 2: https://keeb.dev/2025/09/29/claude-sonnet-4.5-system-initiat...

Turns out local models are pretty great at it, who knew?

Running it in production for free and complaining about the offering is a choice.

They are not complaining about the price, but about the closed source nature of the product.

I'm not running Traefik in production. The features that I need are all closed source so I moved on.

Sure, it's a choice but I think it's more that don't pretend you are open source when your carefully hide things behind closed sourced paid licenses. Be like Microsoft, we have eval version but if you want to use our Windows Server, you will be paying up. Cool, I can make a decision about your software with that in mind.

A choice is a far cry from the "standard" the title purports.

I pass my server motd / startup through a random tte[1] and it makes me happy every time. Thank you!

1: https://keeb.dev/static/login.mp4

Awesome, glad you like it. The next release 0.13.0 will have a 'random_effect' option so you're not forced to handle it via shell function. It also significantly reduces the data being transferred and the CPU usage by adjusting the framerate to a more appropriate 60fps. It should perform much better on low end devices which are typically accessed via SSH such as raspberry pi's.

It says so in the article. Isp's who ignore authorities and allow anything to happen on their networks.

So happy to get this out into the world - we're here to answer questions if you've got 'em.

While the attack vector is completely obvious when you think about it, the gumption to do it is novel. Of course this is the best way to exfiltrate data, it's on a blessed path and no one will really bat an eye. Let's see how corporate-mandated anti virus deal with this!

How can an antivirus even prevent this?

Just needs to prevent the system from booting, like CrowdStrike did

It can't

Even using Claude Code vs. something like Crush yields drastically different results. Same model, same prompt, same cost... the agent is a huge differentiator, which surprised me.

I totally agree that the agent is essential, and that right now Claude Code is semi-unanimously the best agent. But agentic tooling is written, not trained (as far as I can tell—someone correct me) so it’s not immediately obvious to me that a third-party couldn’t eventually do it better.

Maybe to answer my own question, LLM developers have one, potentially two advantages over third-party tooling developers: 1) virtually unlimited tokens, zero rate limiting with which to play around with tooling dev. 2) the opportunity to train the network on their own tooling.

The first advantage is theoretically mitigated by insane VC funding, but will probably always be a problem for OSS.

I’m probably overlooking news that the second advantage is where Anthropic is winning right now; I don’t have intuition for where this advantage will change with time.