max8539

It could be about cost: big LLMs don’t work locally like most games, and even with server hosting, the context for a well-detailed and non-textual game would grow very fast, so processing it would become slow and/or expensive. These are the current limits for truly generative games.

As for generated content in regular games, I don’t see an issue if the content is high quality and free of errors. People don’t like low-quality content regardless of who generated it - human or LLM. It’s just that there’s currently more bad content coming from LLMs, that’s all

I’m trying to use the CLI whenever possible - it’s much easier to install and can be used by both me and the agent. For example, gh seems much easier than installing and setting up an MCP server connection, and it’s more human-readable in terms of what the agent is calling and what it’s getting in return.

For other integrations, I first try to find an official or unofficial CLI tool (a wrapper around the API), and only then do I consider using MCP

I was just thinking about it two days ago - how nice it would be to use my local Claude code instead of the limited cloud version to make some ad hoc changes when I have a fresh idea on a hike. And two days later - here we go, a new release

It’s a small model. A 70B or larger model on a chip would be much more impressive and hopefully would have the same speed

People are used to it, yes, but I’m still not sure about the efficiency. Even screen size makes a real difference when you want to see more data on one screen or switch between windows quickly to compare sources

Well, it’s interesting, but who is heavily working with text, which requires a lot of typing, and only has a smartphone? Phones are mostly for consuming. For creating, it’s usually easier and more comfortable to use a device with a keyboard (PC or laptop)

This is crazy! These chips could make high-reasoning models run so fast that they could generate lots of solution variants and automatically choose the best. Or you could have a smart chip in your home lab and run local models - fast, without needing a lot of expensive hardware or electricity

This is crazy! These chips could make high-reasoning models run so fast that they could generate lots of solution variants and automatically choose the best. Or you could have a smart chip in your home lab and run local models - fast, without needing a lot of expensive hardware or electricity

It’s easy then run a web server for a web UI, but it still looks better than a regular CLI

It could be a native mobile app, but in that case you’d need to think about platforms, the App Store, and so on.

A progressive web app could be a nice alternative: just add the website icon to your phone, and you can open it in seconds.

The app could be self-hosted on a home server, and solutions like Tailscale would let you easily access it outside the house as well. A big plus is that you can open it on all your devices and have a unified database across them.

With a web app, you could even keep the TUI aesthetic - just style it like a CLI interface.

Before vibe coding, I was always interested in trying different new things. I’d spend a few days researching and building some prototypes, but very few of them survived and were actually finished, at least in a beta state. Most of them I left non-working, just enough to satisfy my curiosity about the topic before moving on to the next interesting one.

Now, these days, it’s basically enough to use agent programming to handle all the boring parts and deliver a finished project to the public.

LLMs have essentially broken the natural selection of pet projects and allow even bad or not very interesting ideas to survive, ideas that would never have been shown to anyone under the pre-agent development cycle.

So it’s not that LLMs make programming boring, they’ve allowed boring projects to survive. They’ve also boosted the production of non-boring ones, but they’re just rarer in the overall amount of products

Looks nice, I like this TUI aesthetic, but I’m not sure I could use it on a daily basis. A self-hosted app or phone app might be more convenient