ᵉzᕈuzzle

people aren't getting leverage from AI for the same reason they got laid off, looked over for promotion, or failed to manage their team effectively.

@mitchellh Vibe coders don't see the value of branching, isolated dev servers, database migrations, nor do they know anything about tooling. Managing multiple agents operating on shared everything is doomed to fail. CDEs are expensive or slow or crash. Coder+Mux is almost right imo

I’ve been forcing myself to use the GUI agent apps the past week just to learn. One week isn’t enough experience yet, but I think they all miss the mark (so far) and I’m deeply fighting the urge to do one with my own taste. I’m not gonna do it. I’m not gonna do it.

@jarredsumner llm optimized ctags for typescript. basically some dense view of type signatures per file with some flags for including or excluding things (eg exports, functions, interfaces etc) including line number and length.

what large features should I run claude in a loop to try adding to bun constraints: - must have a large test suite with positive & negative tests that verify it works - must have strong evidence of being something people want & would use bun for

@soychotic nothing worse than talking to humans in 2026

Wedding venues be like: >“click here to download our free pricing PDF! We just need all your information :)” >“thanks for submitting! Someone will contact you shortly!” >no pdf.pdf I am enraged

@cursor_ai TPS reports ain't gonna make you prompt better...

We're proposing an open standard for tracing agent conversations to the code they generate. It's interoperable with any coding agent or interface. agent-trace.dev

@eddieddieddiedd sleepin on it

Maybe if we RESPECT in peace instead of rest in peace our society would have no more problems 🤔🤔

@benhylak single copy of git internal state and branch registration makes it much easier to automate anything that requires git commands

i've been using git for a long time and i still don't get why i should use work trees instead of just copying/pasting the folder

@emollick assuming that somewhere in the training data is a close to ideal (minimal) solution for a well stated (read: easily mappable to emergent world model in an llm) problem i think optimizing context for completeness of the local "world" (codebase) is ideal. complexity kills.

Should we rely on serial agentic work (skills) or parallel work (subagents)? Do we want a hierarchy where one agent decides what the others should do? Can agents act as judges to check in on another agent's work? What checkpoints and tests need to be built into agentic prompts?

With the advent of workable AI agents, we are sort of back to square one on prompting/context engineering. Lots of opinions but there is no clear evidence on the right way to prompt agents, how they should be orchestrated together & how to organize and structure very long tasks .

@ChShersh you are implying we should expand the scope to all computer users?

Can’t believe we all used to stress so much about AI replacing engineers, lol

@GergelyOrosz @steipete subtext: most people are terrible at prompting

"I don't like pull requests (PRs) any more. A large chunk code change doesn't tell me much about the intent or why it was done. I now prefer prompt requests. Just share the prompt you ran / want to run. If I think it's good, I'll run it myself and merge it." - @steipete wow

most of the AI coding rhetoric on here seems to be about how to get claude to "successfully" build complex things. easy answer, build simpler things.

@0xzak write smaller specs

Just shipped adversarial-spec, a Claude Code plugin for writing better product specs. The problem: You write a PRD or tech spec, maybe have Claude review it, and ship it. But one model reviewing a doc will miss things. It'll gloss over gaps, accept vague requirements, and let edge cases slide. The fix: Make multiple LLMs argue about it. adversarial-spec sends your document to GPT, Gemini, Grok, or any combination of models you want. They critique it in parallel. Then Claude synthesizes the feedback, adds its own critique, and revises. This loops until every model agrees the spec is solid. What actually happens in practice: requirements that seemed clear get challenged. Missing error handling gets flagged. Security gaps surface. Scope creep gets caught. One model says "what about X?" and another says "the API contract is incomplete" and Claude adds "you haven't defined what happens when Y fails." By the time all models agree, your spec has survived adversarial review from multiple perspectives. Features: - Interview mode: optional deep-dive Q&A before drafting to capture requirements upfront - Early agreement checks: if a model agrees too fast, it gets pressed to prove it actually read the doc - User review period: after consensus, you can request changes or run another cycle - PRD to tech spec flow: finish a PRD, then continue straight into a technical spec based on it - Telegram integration: get notified on your phone, inject feedback from anywhere Works with OpenAI, Google, xAI, Mistral, Groq, Deepseek. Leveraging more models results in stricter convergence. If you're building something and writing specs anyway, this makes them better. Check it out and let me know what you think! github.com/zscole/adversa…

@martyrdison asparagus

i don’t eat/drink the following: - coffee (never had in my life) - alcohol - sparkling water - candy or really any sweets - energy drinks - fried foods - seafood (yes, that means no sushi) - pork - burgers - cookies or sugary pastries (esp. no icing) - soda - chips/snacks

@keysmashbandit leaderboards for daily tasks. toothbrush coverage. food waste. trash can on a scale that sends you a nightly notification of added mass.

This post is about the people on the leaderboard of the website that I'm using to learn Dvorak

Sometimes I wish I didn't know any other human in the world existed, I think it would make a lot of things I do much more tolerable

meet the “PlattenKreisel” a circular vintage-style vinyl storage console

I've been stuck working from home for like 6 years now and I really want to go to an office again.

@LewistheLight1 i like listening to jazz music records and ecm, live in San Francisco (on divis), and have kids. I'm as tech as it gets though unfortunately. lemme know if u change your mind.

Hi, I’m Lewis I moved to San Francisco 15 years ago. Looking to actively avoid anyone in the tech community. A lil about me: - I ❤️ Pat Metheny & all ecm albums - I think Letter From Home is greatest album of all time Message or DM me! Would love to listen to Pat Metheny w you!

@kdaigle incredible. and all it took was a few petabytes of AI slop code making it in to production.

Been at GitHub for nearly 13 years and I'm pretty sure this request is more than 10 years old... 🆕Comment on unchanged lines in any file inside a Pull Request is now available Gotta keep the core of GitHub getting better every day. 🫡

hard to trust these AI infra companies when every single one of their homepage animations lags like shit.

@hardfiled that's the one. wonder book was the other place but now they overprice everything. I'm out in SF now so haven't been digging in Frederick for a while.

@ezpuzzle This is from the Record Exchange of Frederick, Maryland. Is that the one you mean? I understand there’s a lot of shops out there with that name.

People are talking about the most they ever spent on a record, but let me ask you this: what’s the best deal you ever got on a total classic? (That stamp says Feb 2015, in case you were wondering)