I interviewed some years ago for an AI related startup. After looking at the live product, first thing I see is their prod dB credentials and openAI api key publicly send in some requests... Bad actors will be having a lot of fun these days

It's one of the main things I learned when working as tech support and I talked with users all day. Nobody reads anything.

Or maybe those who do read the docs require less tech support.

A totally understandable situation. Most people just want to use technology to accomplish their immediate goal. I'm tech savvy and I lose my mind every time I get distracted by broken/misconfigured technology.

I used to refer to the helpdesk as the reading desk - “Hello, you’re through to the IT Helpdesk, what can i read for you today?”

The reason I used cursor, other than it being paid by my employer, is that it had a pretty good integration between IDE and the agent workflow.

If I want to mostly direct 1 or more agents I go straight to claude code (codex at home.)

But I still want to have a IDE at the end of the day, I do look and review the code. I still need to direct it to fix some things it doesn't do properly and I dont feel like giving up my understanding of the system I work with (despite what the vibe people say) I don't think it will lead to good outcomes or any benefit in the name of speed.

So for me this direction goes against what I find useful in cursor, and entirely seems to look out for the the 10+ agents crowd. Which makes sense, these are the guys spending +200 $ subscriptions and so on. I'll go back to Zed + CC or Codex.

By the way their new interface looks just like the Codex App.

Hey, very nice! Seems like a great way to have LLMs answer questions about the laws more reliably.

Hey! How can I make this about LLMs?

(Many countries' laws are already available online and included in the dataset they're trained on. The project is very cool for humans though.)

My favourite QoL improvement to any shell I use is to improve the history function(Ctlr+R)I personally like https://github.com/cantino/mcfly

There are so many of these "meta" frameworks going around. I have yet to see one that proves in any meaningful way they improve anything. I have a hard time believing they accomplish anything other than burn tokens and poison the context window with too much information. What works best IME is keeping things simple, clear and only providing the essential information for the task at hand, and iterating in manageable slices, rather than trying to one-shot complex tasks. Just Plan, Code and Verify, simple as that.

There was a post from Apenwarr[1] recently that gave it a name: "the AI Developer’s Descent Into Madness", ending with "I need an agent framework. I can have my agent write an agent framework!"

[1]: https://apenwarr.ca/log/20260316

Sounds like a FactoryFactory to me which tells me Java developers have always been mad

From my experience they are motivated by these two issues that you run into when using Claude Code (or similar tool):

1. The LLM is operating on more what you'd call "guidelines" than the rules -- it will mostly make a PR after fixing a bug, but sometimes not. It will mostly run tests after completing a fix, but sometimes not. So there's a sentiment "heck, let's write some prompt that tells it to always run tests after fixing code", etc.

2. You end up running the LLM tool against state that is in GitHub (or RCS du jour). E.g. I open a bug (issue) and type what I found that's wrong, or whatever new feature I want. Then I tell Claude to go look at issue #xx. It runs in the terminal, asks me a bunch of unnecessary permission questions, fixes the bug, then perhaps makes a PR, perhaps I have to ask for that, then I go watch CI status on the PR, come back to the terminal and tell it that CI passed so please merge (or I can ask it to watch CI and review status and merge when ready). After a while you realize that all that process could just be driven from the GitHub UI -- if there was a "have Claude work on this issue" button. No need for the terminal.

After a while many people then realize this often produces worse results by injecting additional noise in context like the overhead of invoking the gh cli and parsing json comments or worse the mcp.

But they get the dopamine loop of keeping the loop alive, flashing colors, high score/token use, and plausible looking outputs — so its easy to deceive oneself into thinking something remarkable was discovered

It's basically .vimrc/.emacs.d of the current age.

These meta-frameworks are useful for the one who set them up but for another person they seem like complete garbage.

I have my own mini framework that marries Claude and Codex. When I see the clangers that Claude by itself produces that Codex catches, I can’t see how I’d ever just let a single agent do its thing.

Once the plan stage is done is it fire-and-forget for you afterwards?

Most of the jobs listed seem something you can do on autopilot, while your mind is doing something else, which would make sense for creative minds.

Brandon Sanderson often says in interviews that "laying bricks" is the best job a writer can have. He also says being a software engineer is particularly bad job for writers because you cannot do it on autopilot. I can confirm.

Back then, all jobs moved at a much slower pace. There was a lot more off time during work hours.

I gotta try this!

I think discord has been terrible for the internet. A lot of open information has become gated. And now it's gated behind a platform that many of us are not willing to use anymore. Let's hope this pushes out people and communities back to forums and such, but in reality other platform will take over.

My all time favorite console. I keep coming back to it. This to me is a fantastic way to preserve gaming history.