🔥Telegram İfşa

So, basically when working on frontend only use Claude 4.5. Backend, refactoring, logic issues, Codex-High. Hope this helps!

📢 Time for an update on my workflow. This one's a 23 min read, so buckle up. 100% organic and hand-written, like an animal. steipete.me/posts/just-tal…

A/B testing two different approaches to AI code review at work They both respond with “you’re absolutely right” to a GitHub comment I hate it here 😂

Introducing SWE-grep and SWE-grep-mini: Cognition’s model family for fast agentic search at >2,800 TPS. Surface the right files to your coding agent 20x faster. Now rolling out gradually to Windsurf users via the Fast Context subagent – or try it in our new playground!

When I started working on the AI SDK in 2023 I thought it was a long shot to ever catch up with LangChain. Times change.

oxfmt (the oxc-based formatter) is getting very close to ready! - 2-3x faster than Biome - 45x faster than Prettier - Better prettier conformance than Biome means switching won’t cause huge diffs - More flexible line wrapping than Prettier We have tested oxfmt against several

Hope this can run npm scripts Without feedback loops all coding agents are garbage

lol ai coding is going to go so hard with stacked prs

How it feels writing code at work lately

correct and the improvement over just this year has been massive. coding agents started working, and outclassing me at most real world programming tasks. scientists and mathematicians all over are saying it did some small unsolved problems for them. we’ve come so far

Every LLM software engineering benchmark is flawed. I’m finding it increasingly hard to not get nerd-sniped into creating my own.

nothing has cured me of so many anxiounesses of life like marriage + kids. I get to truly feel alive because life is no longer about what i want, but about the very real needs of people who depend on me that i love with a love i did not believe i was capable of

words cannot describe how bullish i am on google right now history will remember sundar's steady hand as one that steered google through some dark days TPUs Deepmind's complete non-chalance at OpenAI/Anthropic's shenanigans Waymo Youtube Shorts

If you are pinging someone who’s busier than you, you should have a single well-thought-out specific ask. Only ask for something you truly cannot do on your own, and where you have done everything you possibly can by yourself. For example, if you’re asking someone to send an

Natural Processing mode has made me love taking iPhone pictures again. Left is normal Apple Processing vs Right is Natural Processing mode

*gets up on soap box* With the announcement of this new "code mode" from Anthropic and Cloudflare, I've gotta rant about LLMs, MCP, and tool-calling for a second Let's all remember where this started LLMs were bad at writing JSON So OpenAI asked us to write good JSON schemas

new plumbing notes just dropped the mechanics behind the Fed's coming balance sheet "bazooka" conks.plumbing/p/plumbing-not…

We're introducing Project AELLA, in partnership with LAION & @wyndlabs_ai AELLA is an open-science initiative to make scientific research accessible via structured summaries created by LLMs Available now: - Dataset of 100K summaries - 2 fine-tuned LLMs - 3d visualizer 👇

it's honestly insane how much work it is very specifically asking for exactly what i want. like with gemini pro and codex i can basically get anything i want done without working as long as the set up is good enough. but i still have to know what i want and knowing it is hard