songodongo

I think you’re vastly overestimating how difficult this type of application would be to an LLM. There’s no need to steal another code base…isn’t yours closed source, anyways?

You could probably get 90% of the way there with a prompt that literally just says:

> Create a TUI application for exploring deployed AWS resources. Write it in Rust using the most popular TUI library.

Bad feedback can be more helpful than good and is often the only type of feedback a product gets. And you may not have received that feedback if you didn’t ship. It’s better to get that information early.

The paper says they used a recursive depth of 1. Does that mean subagents or sub-subagents?

There was already an Israeli pundit on Fox News saying that Venezuela harbors Hamas and Hezbollah operatives. My assumption is that they are throwing that out there to garner support from Trump’s “anti-war unless it’s defending Israel” supporters.

Well, he sure is confident in himself with quotes like “you certainly don’t tell a researcher like me what to do” and “I’m a visionary”. Best of luck.

Are you on the Kiro marketing team?

If you can test how often it is being used by having a line in there saying something like “You must start every non-code response with ‘Woohoo!’”

Anthropic is partnered with Palantir though…

Merry Christmas! Christ is King.

Maybe this is what Altman and Ive are working on…

Are you going to post an example of the CLAUDE.md your team uses?

I get this behavior with a lot with most of the premium models (Gemini 3, Opus 4.5). I think it’s somehow more a GitHub Copilot issue than the models.

Good to know I wasn’t the only one thinking “wait a second…” when reading that one and seeing the spread operator being used.