danialhasan

a big risk of the ai era moving so fast, and everybody working so hard (clauding 14 hours a day) is people never taking a step back to see how the world around them is changing. you need to step back, relax, take a few walks, and WRITE to notice things others miss. you just can't notice some things when youre IN it. you need to reflect, and writing is the best tool for straightening out the mind

@getdelve Cooked💀

DELVE - A New Way Coming Soon

@WarMonitors it was an IR missile that came from under the cockpit, which means the pilot couldn't see it... which doesn't make sense since the aircraft is see-through via cameras piped into the pilots HUD, so the pilot must have been able to see the missile trail. but didn't react. oh well

Now that everyone calmed down can we actually begin to ask how tf Iran was able to even lock on to F-35 in the first place, let alone hit it without flare dumping. Crazy.

@danshipper Great. now run this experiment 100 more times and analyze the results I love science

i asked composer 2 to optimize my production QA process and pitted it against gpt-5.4 composer 2's response won (as judged by both 5.4 and opus 4.6):

Delve just scammed hella companies in the space, if you or someone you know was involved tell em to check themselves!!

heads up that if you use @getdelve for compliance they're blatant fraudsters and have opened you up to massive legal liability. @karunkaushik_ and @kocalars i hope you both go to jail :)

@gmoneyNFT what are the failure modes?

i gave it a shot, but can't do this anymore. hermes sucks ass. all these agents suck ass. they just stop working all the time and then take forever to debug. sticking to claude code and codex in terminal. far and away better than messing with this productivity porn

opencode 1.3.0 will no longer autoload the claude max plugin we did our best to convince anthropic to support developer choice but they sent lawyers it's your right to access services however you wish but it is also their right to block whoever they want we can't maintain an official plugin so it's been removed from github and marked deprecated on npm appreciate our partners at openai, github and gitlab who are going the other direction and supporting developer freedom

You either take the risk or end up working for someone who did. Worst case scenario: you learn. Best case scenario: it changes your whole life and trajectory. Playing it safe never built anything worth remembering.

@yashhsm Very cool article

ABSTRACTIONS[dot]MD

@rivet_dev ohhhh this is interesting

Introducing the Secure Exec SDK Secure Node.js execution without a sandbox ⚡ 17.9 ms coldstart, 3.4 MB mem, 56x cheaper 📦 Just a library – supports Node.js, Bun, & browsers 🔐 Powered by the same tech as Cloudflare Workers $ 𝚗𝚙𝚖 𝚒𝚗𝚜𝚝𝚊𝚕𝚕 𝚜𝚎𝚌𝚞𝚛𝚎-𝚎𝚡𝚎𝚌

I Live for a good problem landscape

the kanban isnt where the value is at, its the provenance of state changes over time that translate into real work you can deploy, evaluate, and improve over time. theyre just an abstraction; and you should remove abstractions when theyre not helpful. at scale, cloud inference on hundreds/thousands of task will just need to be controlled in a different way.

100% of dev is going to be done in sandboxes in the cloud, controlled by kanban boards. Trust me, I love my local machine and gorgeous mac apps, but all of it is just a terrible form factor for running a team of agents effectively.

@badlogicgames dax w the fax

can't believe david cramer and dax are my spirit animals now. but here we are. 2026. wtf.

@danshipper codex had subagents before today, just didnt sho wup in the ui

guess when i started using codex subagents to resolve my production issues lol

@tarunsachdeva passionfruit

This is what I live for, these 2 random weeks before spring that Costco decides to sell fresh passionfruit straight from Zimbabwe

@ryancarson @linear We're building this at trysquad.ai and would love to hear more about you on this

The company that actually builds the agent-first code factory is going to be worth hundreds of billions of dollars. No one has cracked it yet. It can't be the model labs because then you're tied to one model. I'm hoping a company like @linear will do this. I'd happily pay thousands of dollars a month for that (+ the token cost). Basically, we need SDLC 2.0 for the agent age. (Also, the right solution can't rely on gh - we need whoever does this to completely replace it as well.)

@tunahorse21 I must fix this

ive been wondering why, over the last few months at work, ive been seeing more & more slop from jrs not the usual slop either. now its slop with extra layers of complexity wrapped around it to hide it ofc juniors are always going to make mistakes. thats normal. but before, the blast zone was limited by how fast they could open prs. now they can open prs way faster so we keep adding stricter & stricter guardrails but instead of tightening up the implementation, the agent seems to push them into a different pattern. more structure, more branches, more splitting, more abstraction, just because it just decomposes harder. breaks things into more pieces. adds more layers for no real reason during training i hammer home the architecture, and a lot of the time we end up answering it manually anyway because the overhead starts to outweigh the benefit concerning

@mattshumer_ ytou can also say "explain like im 15" because "explain like im 5" abstracts too much so 15 is the sweet spot for quick comprehnsion

Add this to your Codex custom instructions for a way better experience: "When communicating your results back to me, explain what you did and what happened in plain, clear English. Avoid jargon, technical implementation details, and code-speak in your final responses. Write as if you're explaining to a smart person who isn't looking at the code. Your actual work (how you think, plan, write code, debug, and solve problems) should stay fully technical and rigorous. This only applies to how you talk to me about it. Before reporting back to me, if at all possible, verify your own work. Don't just write code and assume it's done. Actually test it using the tools available to you. If possible, run it, check the output, and confirm it does what was asked. If you're building something visual like a web app, view the pages, click through the flows, and check that things render and behave correctly. If you're writing a script, run it against real or representative input and inspect the results. If there are edge cases you can simulate, try them. Define finishing criteria for yourself before you start: what does "done" look like for this task? Use that as your checklist before you come back to me. If something fails or looks off, fix it and re-test. Don't just flag it and hand it back. The goal is to keep me out of the loop on iteration. I want to receive finished, working results, not a first draft that needs me to spot-check it. Only come back to me when you've confirmed things work, or when you've genuinely hit a wall that requires my input."