Serhii Vasylenko

@GergelyOrosz Yesssss… to fight that, I use a hook that scans every input from me to check for specific words and phrases such as “how about” or “btw, let’s” — to nudge Claude be extra sensitive to planned vs creeping scope, and flag that. And that helps.

That feeling of: "I'm in the middle of the code... oh, this is such a nasty hack. OK, let me clean it up as I go. [2 hours pass] OK, it's done, now let me get back to where I was." It just never happens as organically as I use AI agents. I no longer spot stuff as I don't "live in" the code...

@steipete @Cloudflare Are you approaching running Openclaw on their containers infra, or is it something you already figured out?

Still limited by compute, so I built a thing that runs codex in the cloud, powered by @Cloudflare firecracker boxes (and since that's not beefy enough for larger projects, tests are run via crabbox) Uses Ghostty ofc, via WebAssembly. Codex replicated itself, basically.

Found this on Claude Cowork landing page and now I wonder whether it's an easter-egg-thing or really just a placeholder they forgot to remove?.. "Lorem ipsum dolor sit amet, consectetur adipiscing elit suspendisse varius." claude.com/product/cowork

@GergelyOrosz @SamMorrowDrums Yes, but… In situation 2, dev A built a prototype fast while dev B was keep finding good arguments and shipped nothing. And happens too often to be just an ironic anecdote.

Situation 1: dev A thinks approach X is correct, dev B thinks Y is the right way. They argue and try to convince each other. Situation 2: dev A thinks approach X is correct, tells the LLM to implement it. There is SO MUCH learning in Situation 1, lost when using LLMs....

@slobkebap Also in the airport. I am eager to see billboards along the roads with SaaS ads now :-D

are SaaS ads finally starting to become mainstream in Berlin? i can't 😂

@ClaudeDevs Ka-Ching! 🤑

Fast mode for Claude Opus 4.7 is now available in research preview on the API and in Claude Code.

@GergelyOrosz Well, Yahoo search also works reliably, and Bing… 😃 I mean, maybe it is not that SRE made Google’s the outstanding one.

I'm not talking about SRE for other Google products, or across the industry. But SRE for Google Search. Google Search is probably the single most reliable public facing service, globally - the last 20 years.

We like to say there is "no silver bullet" in software engineering: a technology or management technique that by itself promises an order of magnitude improvement in productivity, in reliability, in simplicity." So tell, me what was SRE for Google Search from 2003, if not this?

@GergelyOrosz For me, AI-assisted writing works well in most cases of long-form writing, when a process matters. I have a setup that interviews me and keeps updating the draft (i.e., ghostwriter), but then full stop, and I take it from there. I also tried for socials -- waste of tokens XD

AI - in theory - should make writing easier, thus expressing ideas should be a lot easier. And yet, I don't see all that much more things that are worthwhile to read. Sure, there's a lot more junk. But I don't see more interesting eng blogs, personal tech-related blogs etc. What is going on? (Is this a discoverability issue? Or are people / teams not writing/sharing all that much more, indeed?)

Part 2 — layers made it universal, harnesses made it run: devdosvid.blog/2026/04/07/lay…

Part 1 — flipping the axis to a layer-based approach: devdosvid.blog/2026/03/05/fli…

In Q3 2025, I had an infrastructure migration ahead of me that didn't fit the usual playbook, and I spent the early part of it asking the wrong question about AI agents.

@trq212 As a human, I love the design of those examples. As an engineer, I think MD is better suited for machines as it carries pure info value vs HTML with its tags. And asking agents to put git diff into HTML goes against the harness principles and gives them +1 chance to mess up.

HTML is the new markdown. I've stopped writing markdown files for almost everything and switched to using Claude Code to generate HTML for me. This is why.

My TIL today: making F4 work in Tmux like it does on Mac, and showing me the windows. This requires enabling F-keys to work as standard function keys in macOS settings first. And then this feels sooooo gooooood :-D

@GergelyOrosz I think it is similar to Google investing in Anthropic — the win for xAI is a contract with a customer that grows like a hell. Both companies share some major investors too. If investors see

So let me get this right: 1. Anthropic bans xAI from using Claude (to stop them from perhaps distilling Claude for their own model) (...) 2. xAI gives up ~a quarter of its DC capacity for Anthropic to rent and run Claude A win for Anthropic no doubt. What's in it for xAI tho?

@steipete This is impressive! But will you eventually start killing your apps, like Google? :-D

Me and codex were busy. 🔊 sonoscli.sh — Sonos 🗃️ wacli.sh — WhatsApp 🪶 birdclaw.sh — X archive 🧰 gitcrawl.sh — GitHub archive 🛰️ discrawl.sh — Discord archive 🎧 spogo.sh — Spotify 💬 imsg.sh — iMessage 🧳 mcporter.sh — MCP to CLI 🗣️ sag.sh — ElevenLabs voice 🧿 askoracle.sh — second opinion Upgrading the 🦞 OpenClaw army.