Zelys - DFK Helper

Don't settle: do it all.

Day 10 of The Great Pausening. We've reached double digits and are rooting vehemently for not reaching triple digit days of downtime for our favorite game. Hope is not a honking strategy but will remain hopeful nonetheless.

@rushicrypto Billions of Dollars in spit, duct tape, and courage

Serious question: If we were able to go to the Moon and land there in the 1960s, what makes it so difficult to do it again in 2026? Can’t we just reuse the technology from the 1960s and 1970s?

@nbaschez Just give in and trust it, after telling it to loop several times. You can always just go back.

My biggest challenge with vibe coding / agentic engineering lately has been getting stuck in what I call a "plan doom loop" - have AI write a plan - review myself, seems good - have AI review plan, it always finds something - repeat It drains my time and energy to determine how important the "findings" really are Who has solved this

@DeryaTR_ On a low plan, don't use Opus for anything other than planning. Direct it to use Haiku spawns for research, exploring, and simple tasks.

After 15 min of asking Claude Code Opus 4.6 to continue where it left off. which had hit the limit, 90% of the limit was already used. A few minutes later, it was at 100%. Result: zero tasks completed! This is as bad as it gets. Clearly, the token-eater bug is still not fixed!

@MaMoMVPY You must not be using it much. IQ in the highest 2%, and Claude is beyond me in all respects.

This is again complete and utter bullish. LLMs are nowhere near human intelligence. And Amodei knows this. The question is why do he keep lying about it?

@sandislonjsak Dedicate enough time, and that part of the brain that did code flow slowly makes it easier to multi agent. Multitasking used to stress me out, but no longer.

My brain simply can't run more than 3 agents in parallel and QA all of their work. I am sure I am not the only one. How do people manage 10 at once? Or they simply lie?

@MoeCanDoIt Always run it with custom skills that control output.

I'm on Claude's $200 plan, and for some reason it's getting dumber. I'm talking simple contract review stuff where it COMPLETELY misses important details. Did it get nerfed? What's happening?

Tesla FSD

@allie__voss Just a relaxed metal detector. And, before that, the air white and thick with cigarette smoke.

As someone who never experienced airports before 9/11...was there truly NO security? Like you could just walk wherever like a train station?

@dansemperepico Also way more powerful.

Guys don’t be scared of using Claude Code in terminal. It is 95% the same as the desktop app. You type words, the AI does its thing, and you get words back. But you’ll feel like a hacker using Terminal.

@bitcoinpanda69 Probably what school should be teaching all long.

My strongest blackpill belief is that UBI will lead to an overwhelming amount of mental illness most people if they are honest w themselves need external boundaries and challenges imposed on them You're asking everyone on the planet to self-govern their whole lives when most cant even go to bed on time now when they have work early

@Belly0x The industry, some altcoins, or the OG, BTC?

Be honest. Do you think crypto can bounce back?

@elonmusk Probably the big man saying thanks for the campaign donation, now go away so I can run the printing press.

The fraudsters always come up with the most sympathetic excuses and do so very loudly, because that is what swindlers do

Nice and clean status line for token usage and peak / off-peak time: .claude/statusline.js) 1 let d = ''; 2 process.stdin.on('data', c => d += c); 3 process.stdin.on('end', () => { 4 + let tok = ''; 5 + try { 6 + const j = JSON.parse(d); 7 + const n = j?.context_window?.total_input_tokens || 0; 8 + if (n >= 1000) tok = Math.floor(n / 1000) + 'K'; 9 + else if (n > 0) tok = '<1K'; 10 + } catch (e) {} 11 + 12 const h = new Date().getUTCHours(); 13 const peak = h >= 8 && h < 22 14 ? '\x1b[1;33mPEAK\x1b[0m' 15 : '\x1b[1;32m\uD83C\uDF19 OFF-PEAK\x1b[0m'; 16 9 - process.stdout.write(peak); 17 + if (tok) { 18 + process.stdout.write('\x1b[0;36m' + tok + '\x1b[0m \x1b[0;37m|\x1b[0m ' + peak); 19 + } else { 20 + process.stdout.write(peak); 21 + } 22 });

@elonmusk And no one asks Y.

Model Y is #1

@basedlayer A highly profitable trade of the moment will tend to produce higher gains than passive digital gold. One is a risky investment, the other a store of value.

Bitcoin miners are selling their bitcoin to buy AI infrastructure Marathon, Riot, Hut 8, Core Scientific. The companies that literally produce bitcoin are deciding that AI compute is a better use of their capital than holding the bitcoin they mine Don't think everyone's really sitting with how strange this is. These are the people CLOSEST to bitcoin's monetary properties. They run the machines and understand the halving math. They know the supply schedule better than anyone and they're choosing to sell. The margins on AI hosting are so good right now that the opportunity cost of holding bitcoin is actually negative for them. Think about that for a sec: It's CHEAPER to sell BTC and build data centers than to hold BTC and mine more of it Wild times right now.

@BraddrofliT Even as an atheist.

Serious question: If Jesus showed up today preaching compassion, forgiveness, and helping the poor… how long before he gets labeled a “woke liberal”?

@CAgovernor The Dollar bill signature.

The rest of the Epstein Files? An actual plan for your war with Iran? Something to lower the prices you've sent skyrocketing? Resignations?

@QuetzalThoughts He did say two steps forward for every one step back.

Super weird to think that Trumpism has so fundamentally reshaped American politics whereas Obama feels almost irrelevant at this point. Not at all what someone still in HS when Obama got into office would have ever predicted. Constantly shocked to contemplate it

@AlemTuzlak Facts. I learn something every day.

I find it funny how everyone compares AI to junior devs, I have worked with hundreds of devs across companies of all seniorities. I would say that Claude opus 4.5+ or codex 5.2+ are solid medior-senior devs far surpassing juniors and mediors in code quality. Also, the narrative that junior/medior/senior devs ship some non-slop code that's hand-crafted, home-grown bug free code is even funnier to me. I've seen far too much slop through my career to dare to offend AI output and say its worse than what I've seen. Am I living in some sort of bubble or are we drastically overestimating our own skills to write bug free software?