It's such a dumb discussion.

MCP is an API with some description. It adds tools to your agent, along with some context.

The (common) complaint is that the principle of progressive disclosure isn't working because all tools, with all their descriptions, are loaded into context right at the start. This is a somewhat reasonable complaint, as the structure makes it hard for the harness to progressively disclose the tools.

This is a fundamental issue with anything that just adds a bunch of tools, whether it be via MCP or HTTP (still sad that MCP won over OpenAI's HTTP based approach).

How might it be solved? Well, we could work with sets of tools. That's pretty much what the CLI approach does: Wait until you need it, then invoke the help command to discover what to do exactly. The caveat of the CLI being that it's a nightmare to secure.

At the end of the day, every capability eats some amount of context because the LLM needs to know when to invoke it.

Where can I find that agent harness? A look at their Docs and asking Gemini yielded no results.

Edit: Is it this? https://github.com/Liquid4All/cookbook/tree/main/examples/lo...

FYI: Opencode is very well tuned for Qwen models, but I haven’t found it that rare for niche models to perform badly in it.

So they’re just kind of implying a relationship between the 2 things?

Maybe there is one, but it doesn’t support the underlying “and that must mean AI bad” hypothesis as much as the author may think.

Somebody on the Rsync team has a new tool. They may have neglected their traditional responsibilities using it, but that’s not really a fault of the tool.

I agree that it is not a fault of the tool, but of the human who must have used it improperly.

However, rsync is one of those applications where correctness has an extreme importance. If it fails completely, that is still not so bad, but any kind of subtle corruption in file data or in file metadata can be catastrophic.

I expect from an rsync developer a much higher standard for program correctness verification than for most other computer applications, so these events are very worrisome.

I do not care whether someone uses an AI tool, but I care very much about whether any written code, regardless of its author, is verified very thoroughly, or not.

> Maybe there is one, but it doesn’t support the underlying “and that must mean AI bad” hypothesis as much as the author may think.

It's a tweet. Do you expect thorough null-hypothesis validation from a tweet?

My guess is just open source maintainers trying out new genAI tools out of curiosity. Unintentional slopification

> Why on earth would AI labs be bragging about how little the product they sell actually costs them to make?

Investor confidence. They have a bit of a need for cash (also an interesting part of the profitability discussion of course).

> Also, inference costs are bound to go way down with more optimized architectures

I agree. Jimmy is incredible, I wonder what non-toy use cases they have. Surely they’ll come out with updated chips soon.

That said, I was apparently a bit over-excited for Groq and Cerebras. I thought they’d quickly dethrone Nvidia for inference, but not so far. Even the GPT spark trial isn’t seeming to go far.

Interesting to see that insider trading is considered illegal after all.

When will the white house insiders see the same fate?

Well, you see, the house always wins

In fascism, laws are for binding the out groups.

The odds have shown to be largely correct, thanks to people profitably arbitraging away inaccuracies.

It’s not rare nowadays that speculation on some topic will include the Polymarket rates. Google searches: Maybe not. Maybe that’s just gambling for the fun of it.

Congratulations (I guess), but what does “independent” mean here? Who bought it from whom (and why)? Is it employee owned now? Is it transitioning to a foundation?

OpenCode does show them when you select so in the settings - at least I’ve been getting very long traces so I’d be surprised to learn they are summaries.

Very nice, thank you for this! I’ve been wishing for a way to control when the wallpaper freezes. Can your app also keep it playing on the lock screen?

this is stated in the post my dude. not even at the github link, just like, a few paragraphs in.

> the video keeps playing on the desktop (not just the lock screen)

My dude, that doesn’t really answer the question unless you infer more than people usually mean. This is about changing the behavior of the video stopping after considerable time, the default behavior is that it keeps playing for a good while (while on the desktop it stops close to immediately), which I assume is what they are referring to by “just the lock screen”.