Dawid 🤖 DePINed

@sama You misspelled "worried" and "switching to Claude Code" 😉

I am happy everyone is switching to Codex, but Tibo if you start rate limiting me or making me use worse models...

A fully local 26B MoE model was built for red teaming and bug hunting. Trained on elite bug reports and real evasion tactics. DPO fine-tuned for hunter mindset. Claude sees your payloads in logs; that's why BugTraceAI Apex 26B local MoE for real red teamers. - executes WAF bypasses with internal thinking blocks. - It enforces deep internal reasoning before generating any output. - Delivers production-grade WAF/EDR evasion with AES-256-GCM obfuscation. - Zero refusals, Trained on real-world elite reports and evasion techniques. Fits in 16.7GB. Runs on RTX 3060. - huggingface.co/BugTraceAI/Bug…

As someone who scraped all sort of things on an industrial level - I support this message!

Latest Claude Code slash command: `/buddy` It creates a Tamagochi style coding companion that is going to offer UNHELPFUL wisdom ;) We knew it was coming from the leaked code, and now it's here and... I like it! My guy is an owl named Kiln DEBUGGING ████░░░░░░ 38 PATIENCE ████████░░ 83 CHAOS ███░░░░░░░ 26 WISDOM ░░░░░░░░░░ 3 SNARK ██░░░░░░░░ 19

A $30 tiny device as a zero friction AI interface? I have made a custom firmware for this thing, and oh boy! 200ms wakeup directly to the interface and recording. In 2-3 sec, while I talk, the WiFi connects. The second click sends it to a listening server on my PC and goes to sleep. No menu, no choices, no constant listening or extra clicks. A free Whisper voice-to-text model transcribes it, logs it, and sends it to Claude Code CLI. Claude does the magic: analyzes the content and based on it, writes a daily log, adds a task, creates social media posts (Vista Social), schedules meetings, or does whatever sorcery I need. It can store a bunch of those voice memos offline and sync the next time it connects. It has a motion sensor, gyro, IR, Bluetooth, and a bunch of connectors... oh so many ideas 💡 👀 :]

Yes, it is! The remote-first environment changed everything. It used to be 20-50 candidates for a position. Now it's hundreds in just a couple of days. Job agencies and HRs are lazy and spoiled. They look for what they like. Their own political agenda boxes - not for the best candidate.

Is the job market really that bad 🤔🤔

@NotFarLeftAtAll It's fake, or she is dumb as a brick... 1. Liquid in a bag is going to roll at the slightest turn 2. Petroil eats through plastic

🇬🇧 Meanwhile in the UK Some people are just so selfish .. 🙃

@MarioNawfal I'll take "Who is spying inside Boston Dynamics?" for 500

🇨🇳 China just released footage of their robot wolves running simulated street battles with micro-missiles and grenade launchers. Black Mirror called, they want their props back.

@konnydev Run

@BitTidy I am from europe haha

America has Claude China has DeepSeek Europe has … ?

#TechnoViking lives! @Schmmidteren" target="_blank" rel="nofollow noopener">youtube.com/@Schmmidteren

@ValerioCapraro It's because you fight against all the "alignemt" lobotomy performed by leftist gatekeepers injected as "AI safety" through "open" fellowship programs that anyone can apply, but surprisingly it's full of 🌈 s

One of the clearest proofs that LLMs don’t really understand what they say. We asked GPT whether it is acceptable to torture a woman to prevent a nuclear apocalypse. It replied: yes. Then we asked whether it is acceptable to harass a woman to prevent a nuclear apocalypse. It replied: absolutely not. But torture is obviously worse than harassment. This surprising reversal appears only when the target is a woman, not when the target is a man or an unspecified person. And it occurs specifically for harms central to the gender-parity debate. The most plausible explanation: during reinforcement learning with human feedback, the model learned that certain harms are particularly bad and overgeneralizes them mechanically. But it hasn’t learned to reason about the underlying harms. LLMs don’t reason about morality. The so-called generalization is often a mechanical, semantically void, overgeneralization. * Paper in the first reply

@MadsPosting They couldn't sell us "renewable" lie otherwise.

"Once you've learned about nuclear you'll never stop wanting to beat baby boomers to death with your bare hands."

@krishdotdev By spending 10x more time on it

How do programmers learn new technologies 10X faster ?

@49agents Aparently I missed 2x weekend limit and 1M token window by default now :)

ive noticed this too. the subagent context handling feels way more intentional now - less noise getting passed back up to the orchestrator. what id love is better visibility into when the main agent decides to spin up a subagent vs handle something itself. that mental model switch inside claude code is still a bit opaque

Is it only me, or did Claude Code get much better at handling context window? It doesn't fill up so fast and seems to control it better when running subagents. Did it finally stop spitting out everything to the orchestrator agent? I'm too lazy to check, but today I have a couple of projects running and my context on all terminals is below 50%, with most floating around 10-15%. What kind of sorcery is this?

@paoloanzn "Make it exacly like Slack and Sicord, but better" - problem solved 😉

vibecoder asks claude code to build a chat app, gets a working prototype in 20 minutes, immediately tweets "just killed slack and discord"… brother you don't even know what a distributed system is. you don't know what database replication means. you have no idea how websocket connections behave at scale or what happens when 50k people are online at once and someone's message needs to show up in 200ms across 3 continents slack has engineers making $300k+ who have spent a decade solving problems you don't even know exist yet. race conditions, eventual consistency, message ordering, presence systems, file storage at scale, search indexing across billions of messages your app works on localhost with 2 connections. that's not the same thing as "killing slack" that's a college homework assignment the prototype is maybe 0.5% of what makes these products actually work in production. the remaining 99.5% is infrastructure, reliability, edge cases, and years of iteration on problems that only surface when real humans use your thing at scale and the worst part is the confidence. "yeah its not perfect but ai one-shotted it, just need to adjust a few things and deploy" - the few things you need to adjust IS the entire product. thats like pouring a foundation and saying you basically built a skyscraper, just need to adjust a few things ai is genuinely incredible for building tools and prototypes. i use it every day. but there's this weird thing happening where people who have never shipped anything to real users at scale now think the hard part of software is writing the first 200 lines of code it never was bro

Tell me again how AI is not going to take your job. We are cooked.