Atotalnoob

Samsung TVs (and probably others) will connect to other Samsung TVs in range and exfiltrate data via that route… merely not connecting the TV doesn’t present data harvesting

Iirc, Amazon uses a tool they call “sidewalk” for this purpose

It’s super easy to self host ci, I hosted one with a hetzner VM (pick one in the cheapest region, it doesn’t matter much where CI lives

At my company we explicitly tell candidates what parts they can do with and without AI.

It’s not hypocritical like you seem to be implying. It’s seeing you still have a brain and can do the job with or without the tools. Otherwise we would just prompt AI ourselves…

Then go change them?

Don’t worry, you can use react for TUIs in the terminal.

Claude code uses react, iirc

Sqlite?

The USAF thing was a thought experiment, nothing based in actual reality.

Net neutrality would have prevented this kind of pay to win

Where are you all finding these for sale? I couldn’t find anything other than “kits” or on eBay for $400

Dotnet minimal APIs plus mssql is fine for simple crud apps… I would do Postgres, but that’s me.

Swapping mssql to SQLite would also work perfectly

Consider underclocking/undervolt it a little, especially if you aren’t using the GPU to its full potential.

It will last longer and hopefully the GPU prices will recover in a few years.

I hope

You are replying to a comment about AI slop with an AI generated comment? Bold move.

All of your comments seem to be AI generated.

Google 100% was wardriving and all of your other items.

However, you can increase gps accuracy using wifi. GPS is not that precise (part of which is government regulations) and WiFi does help immensely with accuracy.

Skills and MCPs CAN bloat the context window. Some harnesses do more progressive loading of skills/mcp depending on how many you have. Check your harnesses docs.

Personally, I feel skills+CLI is better, since only the description of the skill enters the context window until it’s required and CLIs should support help flags which will allow you do have progressive discovery.

And a CLI is also easy to use yourself.

Meat is already tracked to its origin. They can track a specific box to the farmer and supplier.

Anytime there would be a recall, they would include the lot number and manufacturer dates.

Source: I worked in meat packing and boutique butcher shops.

Edit: even things like ground beef from trimmings, we would write down every lot that went into it

I don’t get the point of these.

Ghostty has tabs which show the icons for the terminal TUI. You can just look at those for the status.

Same with windows terminal, and you can use either of those to have tiling.

Yes, but it’s slower on non-bun runtimes. Nowadays, I believe they bundle bun with CC

They chose to use a react rendering to native TUI renderer, which was a source of a lot of performance issues.

AFAIK, they have written a new renderer.

There is randomness in LLMs. Both papers authors probably ran the bench 1-N times. Depending on that, they might select an average, max, least, etc. They might also have discarded outliers.

Like the other person said 5% variation is probably expected

Why can’t you use the charp LSP plugin for Claude code?

What does this give beyond that?

An academy award for software?

The color of the text and background of the Linux button contrast poorly in my opinion. It made it a bit hard to read.

This is just insurance, dude

12k for insurance is not enough.

Normal commercial insurance is much more expensive, let alone ambulances.

And don’t forget, the EMS employees need malpractice insurance and the company needs liability and workman’s comp insurance above the normal commercial levels

You need more than the crew in the ambulance.

Dispatch, maintenance, janitors, HR, managers.

They also need continuous training, and that is also a large salary cost for training staff (you have to pay them and the trainer).

Uncertainty is one thing, but a high chance means it’s 51% or higher to me.

Based on that, the bun rewrite messaging was fairly misleading.

Doesn’t that kill your token caching changing models midstream?

Add hooks to block it from being able to edit the files you don’t want changed.

You’d never be able to update it’s knowledge.

LLMs need retraining to incorporate new knowledge.

Baking them into wafers means they will be out of date by the time they finish the first wafers.

A million tokens is like $15 with SOTA models… that’s their allocation?

Sometimes the best option is to enhance observability and see when it happens again