Dmitry Sobolev @sobolev

@petergyang @Scobleizer We need Fable 5.2 version and weekly limits reset

x.com/i/article/2092…

"Playing it safe isn't actually safe. You're just trading the chance of winning for the guarantee of nothing."

Chris Lakin @chrislakin

a week ago

what your risk-aversion costs you

126 1K 9K 3.3M 5K

OpenWorker -- an open source agent that doesn't just chat but completes tasks on your laptop -- just released a new version with many features for security workflows. After our initial release, many users found it especially useful for cybersecurity. Attackers are already using AI; OpenWorker is committed to giving defenders the same leverage. Running an agent requires both (i) A model and (ii) A harness (the software around the model). Because the OpenWorker harness is fully open source, security teams can audit it to make sure we haven't built any backdoors that exfiltrate your code and data to some company or even a foreign adversary. OpenWorker now comes with built-in cybersecurity agents for (i) Scanning your code for vulnerabilities. (ii) Scanning dependencies for supply chain injections. (iii) Checking your cloud security configuration for attack surfaces. This enables developers to do much more security work before deployment (part of what's called the "shift left" movement). You choose the model: you can run open weight models fully locally so sensitive code never leaves your machine. This helps with legitimate security work (like reproducing a known exploit to defend against it) that can trigger refusals in leading closed models. Or use your ChatGPT subscription, or stealth preview models like Ox Alpha, or any model via API key. Thanks also to all the open source contributors! Join work with @rohitcprasad so please follow him too to get more frequent updates. Try it out: openworker.com Code: github.com/andrewyng/open…

Junie can now run entirely on your machine, and it’s completely free to use. One command: /local. We picked and optimized the whole stack around the model: agent harness, model parameters, and inference engine. No tokens, no quota, no code ever leaving your machine.

The workflow I use for LLM coding: - Claude Code in a Docker container with full permissions. I use scripts to sync code in and out of the container, and do git commits myself. - Develop a spec for desired functionality through conversation - Have agent make a plan for implementation, broken down into steps that are as small as possible while still being meaningful - After iterating on the plan, tell agent to implement and to stop after each step - After each step, have agent do a pass finding and fixing test gaps, and then have it do a cleanup pass to simplify/delete code and comments - Review all code, iterating on fixes with agent, before proceeding to next step It's extremely rare that some change isn't needed on every review.

For people who still believe that a language model works by memorizing text: that's just not how they work! And you should be embarrassed that you don't know this by now. To learn how LMs actually work, the standard textbook on the topic is free: web.stanford.edu/~jurafsky/slp3…

Gavin King ⍼⍼⍼ @1ovthafew

a week ago

@patrick_scho @teamloomis Modern training regimes train on two orders of magnitude more text than the information stored in their parameters. So it's simply physically impossible for a model to learn all its training data verbatim. And the information it does learn is stored relatively inefficiently.

3 0 4 16K 2

Apple Pay support is finally coming to Walmart starting next week. Walmart refused to offer Apple Pay for more than a decade. The retailer attempted to push Walmart Pay, its QR code-based payment system, instead. Apple Pay and contactless payment support will begin rolling out to Walmart and Sam's Club starting on August 24, reaching all stores in the United States by the end of this year.

@Kaitarann @facebook Btw have you figured out how to see own listings? )

@Kaitarann @facebook Why do they show Jobs there??? I need to see my listings, not freaking jobs!

@facebook I hate the UI changes for Marketplace. HATE! It’s so hard to figure out how to get to my listings! It’s stupid and unintuitive!!!!!

Today, we open sourced Mojo 🔥. Announced just now during the ModCon keynote, effective immediately, Apache 2.0 License. Thank you to our community for waiting patiently and building alongside us. #ModCon2026 Full blog: modular.com/blog/mojo-open…

From the Director’s Cut…

If you want to buy my "Modern Mathematics" book in PDF or MOBI format, you can do it here: artima.com/shop/modern_ma… (there other good books too, see artima.com/books)

🧩 DeepSeek Harness v0.1 is now available in Developer Preview! 🔹 We’re opening it up to developers building agent harnesses worldwide and open-sourcing the codebase in MIT license. 🔹 Powered by the Cordis meta-framework, DeepSeek Harness is an agent harness built around one core idea: Everything is a plugin. Models, tools, skills, sessions, sandboxes, filesystems, loops, orchestration, and UI are ALL implemented as plugins, and can be mixed, matched, replaced, and extended. Try it now! github.com/deepseek-ai/de…

I'm happy to announce that my book, "Modern Mathematics", is on Amazon now. Kindle version, $28. amazon.com/dp/B0HDML7RCF/… Modern Mathematics is an accessible introduction to the areas of modern math that matter most in software design. 1/

Пробую внедрить github.com/github/spec-kit в нескольких проектах используя GH copilot, Claude Code и Cursor IDE (три разных проекта) Выглядит интересно — писать и валидировать спеки, а не код. spec → plan → tasks → implementation Есть плагины, например, чтобы сразу сделать GH Project github.com/mancioshell/sp… создать тесты github.com/Quratulain-bil… и тд. Вы не пробовали?

Very useful way to make Claude's output really get to the point: github.com/alexgreensh/at…. I find the rundown style quite effective.

"Now I could write at length about all the caveats. All the potential dangers. All the ways it's not quite there yet. But it would be an insult to the marvel of this moment... This is simply the most fun I've ever had with a computer." world.hey.com/dhh/endless-ex…

For anyone with endless ideas, this agent age is nirvana as those ideas are met with endless execution, endless exploration. I've never had has much fun working with computers as I do right now. What a time to be alive.