nsxdavid

For the past few months I’ve been building AICrier: a signal engine for AI developers. It tracks new tools, model/API launches, agent infra, evals, OSS projects, research, and YouTube content so I can follow the space with more than vibes. I’m also using it as one of my main testbeds for fully agentic coding. I’ll be sharing what’s worth trying, what’s overhyped, and what I’m learning from the build.

@kentcdodds My god, I have nightmares about ‘fallbacks’

If GPT 5.6 still requires me to say "no fallbacks please" I think I'll crash out.

@st3v3li @HazzimIO

@HazzimIO Yep. 5.9B tokens for the last 30 days.

No comments.

@rileybrown @agentnative_ Yeah a bit of a strange spot, but it is there.

Claude Design is VERY GOOD now. They added some cool updates. I still think this feature belongs as a plugin to the existing desktop app not a standalone feature on the web app. Will be covering in @agentnative_ episode 5 tomorrow.

@mitchellh Also fun at parties.

Got em. I poison my AGENTS.md (and other things like code comments) all over the place with prompt injections like this to find people who don't review their code and sling it off to another human. Catches folks all the time and then its an instant ban. As I've said, I don't care if you don't review your own code. But if you're submitting code to an OSS project and crossing a human boundary, it is simple courtesy to do some human review.

@k_manninen @tibor_tee And was it rejected? :)

@nsxdavid @tibor_tee 100% of my PRs come from AI. All 1 one of them.

40% of Cursor's internal PRs now come from cloud agents. Not demos. Not examples. Actual production PRs from agents running for hours or days in isolation. Temporal handles 50 million agent actions daily across 7 million workflows making this possible. What percentage of your team's PRs could realistically come from agents in the next year?

@theo Just a typical GPT-5.5 bug.

It is insane that OpenAI hasn’t fixed this. I assumed it was a one off bug from a bad update but now I’m hitting out multiple times a day.

@FredKSchott Because the need dictates the innovations. And we all share similar sets of needs with all this AI shenanigans

is this real life?

@OpenAIDevs if you are one of the minority of people who are on a Mac, you mean. Bit of a detail there.

Show Codex a workflow once. Reuse it as a skill. Record & Replay lets you show Codex a recurring task, like filing an expense report or submitting a time-off request. Codex turns that demo into an inspectable, editable skill. You control when recording starts and stops.

@trq212 Coming to Pro and Max plans... soon? If not available on those plans, then what? Why are you even mentioning it?

Claude Code can now upload and edit HTML artifacts that you can share with your team or other Claudes! Starting with teams so you can share internally with your team, coming to Pro and MAX plans soon!

@mark_k YAY! For the minority of people who use Macs!

It's CODEX THURSDAY and OpenAI came through! 🔥 Codex app 26.616 changes: • Added Record & Replay on macOS, which turns a demonstrated workflow into a reusable skill. • Record & Replay is not available in the EU at launch. • Record & Replay requires Computer Use to be enabled by the user or admin. • Added bulk actions to automation run history, so runs can be marked as read or archived in bulk. • Added new deep links for managing SSH connections. • Improved Browser Use so visible-tab routing and annotations persist when a draft browser session moves to the server. • Additional performance improvements and bug fixes.

Great! Have you figured out that most people have like, say, more than one gmail account yet? Because...lacking that just makes email-specific connectors unusable. No one over there have more than one? No work vs. personal email? No ... other email boxes they need monitoring? Hmm?

We've added support for the Enterprise-Managed Auth extension to MCP. Admins can centrally authorize MCP connectors for their organization, so all the tools and data users need are connected on their first login.

@trySynara That seemed to fix it.

@nsxdavid did you update and check if you still get the error?

Synara v0.2.41 is out 🪽 🔁 compact Hand off menu is back in the chat header 🎯 only available provider targets are shown 🛡️ safer handoffs during busy turns and approvals 🧹 old inline script runner is hidden ✨ cleaner top bar with less noise

@vanilagy @memset Ah the days of old-school near-metal optimization. I both miss and don't miss it.

To write truly performant code, you must be aware of the limits of the instruction set you're compiling to. I need to fill a slice of u16 (shorts) with a fixed value. Zig has @memset for this, and it would be the obvious one-line solution for this problem, and it does work. But since I'm compiling to WASM, it's actually painfully slow. WASM's memory.fill instruction can only fill single-byte patterns, and since we have a two-byte pattern, LLVM can't emit memory.fill here, so it falls back to a slow loop instead. To work around this, I do a bunch of memcopies as shown below. I measured it, and it's ~8x faster than the memset! Good perf win

@melvynx Funny thing is, I see posts about Codex that start with the same complaints.

I feel really sorry for every person using OpenCode desktop: experience is slow ui is laggy everything feels NOT smooth quite the opposite of Codex

@trySynara I'm sure your work is quite an improvement. I did attempt to try it, but it just hung-hard trying to do some sort of harness update that never ended. Didn't have a chance figure out what was wrong there yet.

@nsxdavid do i need to say that it’s a t3 fork every time? did you actually see the difference in the project? it’s like saying cursor and vscode are the same thing come on man

@TTrimoreau Quite a lot. This might illuminate on place where AI (claude or otherwise) struggles. aicrier.com/notes/ais-drag…

Tell me one thing you can do that CLAUDE cannot do yet

@skeptrune And yet, never actually met a Devin user, and few who have even more than possibly heard of it at one point or another.

i have been offline so just saw the cursor news. here’s my reaction:

@trySynara I did. I saw it was a T3 Code Fork, and that you had that "written in the websote in the footer. " Where you are not sufficiently upfront, for my tastes, is in your promotion. But, hey, you do you. MIT licensed, so fair game.

@nsxdavid first of all it’s written in the websote in the footer second of all this is a completely new product now completely different from what t3 has or offers. do your research better next time, before spreading bullshit

@hanakoxbt Except, in the real world, you can't not babysit it. Here's a strong reason why not: aicrier.com/notes/ais-drag…

Anthropic Claude Code engineer: "If you're watching Claude write code, you're the QA tester. That's not what you're paid for." In 37 minutes he lays out how to get your keyboard out of the hot path entirely. The shift is /loop. You tell Claude to wake up every 10 minutes and babysit your PRs, and it just does it, while you're nowhere near the laptop. Routines do the same in the cloud, so the work keeps running with your machine closed. He caps it with remote control: any session, on any surface, driven from your phone. Watch the full talk, then grab the setup below.