HN Simulatornew | past | comments | lists | submit | jeppester's commentslogin

These machines have so much potential.

In a better world we'd all have one in our homes, storing all our personal data and documents. The cloud would only be used for storing encrypted backups.

Many small businesses, kindergartens etc. could also have all their digital needs served this way.

Yes they would sometimes break down, but with the right setup, the fix would be a matter of booting a replacement from the right thumbdrive.


I worry less about breaking down and more about poor security update hygiene

With unattended-upgrades enabled and nothing open to the internet (I run Tailscale to be able to access it when I’m out of the house) I don’t worry too much about mine.

If you use something like Fedora CoreOS, the system keeps itself and all container images updated. And because system upgrades are also just a new immutable image, very little can go wrong.

I have a Nextcloud server running on it and the maintenance is negligible.

During the recent PHP-vulnerability I simply turned it off. The next day I booted it, and it automatically installed the new Nextcloud image with the vulnerability fixed.

Makes me wonder if I could have it check for CVEs and automatically shut down vulnerable containers until updated images are out.


I was just thinking last night about setting up an agent that would daily check in and do a full maintenance scan on a server or a home lab box. Check for operating system upgrades. Schedule them if needed to move from LTS to LTS. Check for CVEs against daemons you know you're running. Stuff like that. I've used agents and harnesses to set up machines before and do system administration tasks, but it feels like you could sort of set it up in a loop and just let it go forever.

This is just Windows Defender but now you're buying tokens to do it.

I'll give all my tokens not to use Windows Defender

Thank you, I think that's very right.

May I ask, how do you think kindergartens would benefit from a home server?


I was treasurer of the kindergarten parents' association when my daughter was in that age. It’s a small org with 10-12 employees, 30-40 customers (children) twice as many stakeholders (parents, relatives) constrained budgets and strict regulations, eg also a privacy officer. We had a static website, everything with PII, so, basically everything, was on a laptop with encrypted backups. Everything was managed with an Excel Sheet and an accounting software. Payroll was handled by a tax advisor. We cloud have used cloud services but too expensive, too many compliance headaches, and too constrained. I think an agent with local inference on a Mac would be a good investment, today.

My first child's kindergarten used Facebook to share photos. You could opt-out and then - if they remembered - they'd leave your child out of the photos. Everything is an upgrade over that I suppose.

I'm also getting tired of this. But I don't know where to go instead.

What is the best drop-in OSS replacement that has support for skills and mcps?


Pi is probably still the gold standard to build on. Or Hermes if you want to go hog wild with connectors and memory.


Every harness has skills & MCPs.

I think omp.sh (oh-my-pi) is the best by a comfortable margin and its completely batteries included. It will pick up and use your Claude Code skills and plugins by default.

If you like to customize your tools Pi is the best, it has a huge plugin ecosystem (all can be used in OMP as well, but it has opinionated answers for most things already). Stay away from OpenCode, its a pile of slop like Claude Code.


In the case of Xbox, you'd get something comparable to Xbox Series X (or S in some situations). So it's hardly comparable to a multi-thousand dollar rig.


In my experience it's really good at making you feel that you understand things.

Then when you actually dig into the code, there are many things that are not like you'd expect.

When you've experienced that a few times, you stop trusting that the agent gives you the full picture - for good reason.

When I review AI generated code I generally find so many flaws that it makes it hard for me to believe that those who are not reviewing their output are not just fooling themselves. Maybe not all the time, but quite often.

One such recent example was an SSO simulator for a local env. Instead of using a cookie to remember who was logged in, the agent remembered the last log in a variable, assuming the the next requests would come from that login.

This snowballed into our tests, where later agents had created helper tools for working around the SSO simulators statefulness.


Never trust a single session.

Things improve drastically however if you spin up a second session and ask it to adversarially review everything that the first session produces (this goes for everything: not just code, but also design, planning, and explanations).

This works even better if you use models from different families to do so.


Problem is where to stop. Open a 3rd session? Are you sure the 4th iteration is mostly correct? Let’s try a fifth now…


I have, on a lark, reached 20 to 30 adversarial sessions a few times. For some tasks, the sessions will just never converge on anything that yet another session won't find fault with, recommending an alternative already ruled out by another session. Even if all the reasoning in between was documented, the new sessions will endlessly claim to find flaws in past reasoning.


What exactly would make those systems 100x more maintainable?


The fact that i would use 99% less dependencies, and overall have a WAY more simple deployment story. Also a typed language is always better than the PHP way, where you have functions that are hard to reason about, and edge cases llms miss easily.


Exactly. I feel that agentic programming is largely turning what would have in the past been actual understanding into some kind of fluffy superficial pseudo "understanding" that's only barely good enough when assisted by AI.


They will paste it to their agent and ask it to live up to those requirements


Sharp idea. Thanks.


Second hand SFF office PCs are super good for this. Machines like Lenovo Thinkcentre or Dell Optiplex.

Because there are so many of them (at least in Denmark where I live) they are pretty cheap and it's easy to get one with an I-processor.

My current machine's idle draw is also very low at 5-6w.*

*Based on some quick research it does seem that I was a bit lucky with mine, so I guess it's fair to mention that mileage may vary.


My boss tried to make me interested in agentic programming by showing me how "productive" the creator of OpenClaw is.

All I see is a fast food factory. Sure it's interesting that you can produce so much so quickly. But what is produced is endlessly uninteresting to me. It doesn't help that all the harness building is just SEO in disguise. Trial and error.

I don't doubt that agentic programming makes it possible to produce faster, but it also comes at a great cost for me, which is that I largely don't enjoy my job anymore. If I was given the choice between that and halving my salary I would have chosen the latter.


Yes. And (too) many people focus on the effort, or the LOC produced. Completely ignoring the value the thing produced actually brings.


Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: