Nick Cooper

@clare_liguori @dsp_ I just like the hardware of macbooks. Ultimately there's a Linux cluster somewhere I'll connect to for most things anyway

@dsp_ @nicknotfun Interesting, I did not have a great experience with Ubuntu + Framework and went back to X1 Carbons

Okay I am think I am convinced to get a MacBook

@embirico heavily heavily use a mix of both. queue is for the next thing i want it to focus on, steer is when i see it's going silly or i think of something very on-theme/change my mind

Do you prefer queuing or steering in Codex? What's your use case for each?

@AmericanAir It's your JFK lounge. And I'm good

@nicknotfun Hi there, when you have a moment, join us in DM with your confirmation trip code and lounge information.

@AmericanAir your lounge bathrooms have the worst UX I've ever experienced. Guess which unlabelled faucet does what!

@vkhosla The art of doing science and engineering is still one of my favorite reads.

Agree it’s very important. Read it years ago and it has stayed with me.

how did Allbirds pivot to AI compute hardware before the shoe company literally called ASICS

My new way to setup an LXC container is to put codex in my core image and I just ask it to do whatever. Today's achievement is it configured my two nics as a bonded group for hot fail over.

@steipete Sure. What's needed?

Anyone here who wants to help with WhatsApp CLI? It needs love, and I can't focus on it right now. github.com/steipete/wacli

@pamelafox I'm very big on extensibility and meta enables a lot of that. All of MCP Apps was only possible by using meta.

MCP server developers, any of you using _meta yet? We're having an internal debate about what kind of data belongs (or doesn't belong) in there. #_meta" target="_blank" rel="nofollow noopener">modelcontextprotocol.io/specification/…

@dsp_ @Poita_ Ah, the classic HMMHRPC

@Poita_ You mean "His Majesty's Most Honourable Remote Procedure Call?"

It’s not everyday that you get to be invited to No. 10

@RhysSullivan @dsp_ the tool fallback isn't too bad an idea as a default behaviour for clients? Something like if the server has them / client doesn't proactively indicate support?

your MCP server should include skills in the resources part of it, and for clients that don't support resources ship a 'load_resources' tool as a fallback

- Drafted a blog post - Used an LLM to meticulously improve the argument over 4 hours. - Wow, feeling great, it’s so convincing! - Fun idea let’s ask it to argue the opposite. - LLM demolishes the entire argument and convinces me that the opposite is in fact true. - lol The LLMs may elicit an opinion when asked but are extremely competent in arguing almost any direction. This is actually super useful as a tool for forming your own opinions, just make sure to ask different directions and be careful with the sycophancy.

@sherwinwu Exactly the same here.

Set Codex to this and don't ever look back

@thsottiaux @jasonkwon I actually quite like it. It also helps me as a reviewer. I'd actually love first class tool attribution.

Do people like this? We don't do this for codex because it exists to help you and it's important that you remain the owner and accountable for your work without AI taking credit. At the same time it does mean that you can't trace how popular codex is among repos.

@notsunsakis @bcherny don't call it vibe coding - that's associated with yolo i smash head on keyboard, not thinking, engineering, building, testing, debugging, iterating. agentic engineering, or just...coding. We move faster, but it's still hard.

I resigned from OpenAI. I care deeply about the Robotics team and the work we built together. This wasn’t an easy call. AI has an important role in national security. But surveillance of Americans without judicial oversight and lethal autonomy without human authorization are lines that deserved more deliberation than they got. This was about principle, not people. I have deep respect for Sam and the team, and I’m proud of what we built together.

GPT-5.4 is launching, available now in the API and Codex and rolling out over the course of the day in ChatGPT. It's much better at knowledge work and web search, and it has native computer use capabilities. You can steer it mid-response, and it supports 1m tokens of context.

always run with xhigh reasoning

@dsp_ @ibuildthecloud I used it to pxe boot and reset an old computer to a Debian+ variant. Not novel but I have no idea how uefi pxe works.

@ibuildthecloud What do you want from your own distro?

So I set out with the goal this year to create a linux distro from scratch for a user of one, me. I thought AI could help me. Current status: I built a tool, to build a tool, to help me use AI to build a distro. So what have I accomplished. Nothing yet. The typical, before I do something I first have to write a new operating system.