IM

Yet another Software Engineer

[Thariq @trq212]

We removed ~80% of the Claude Code system prompt for our newest models, this is what we've learned about writing system prompts, skills and Claude.MDs for them. https://t.co/6DZwSrZjE9

[Lei Chen @chenleidev]

I am the current maintainer of Stylus. The Stylus library has been flagged as malicious lib, which has caused many lib and frameworks that depend on Stylus to fail to install. Please help me retweet this msg in the hope that the npmjs official team will take notice of this issue

[IM @ivandevp]

@FredKSchott ✋

[Matteo Collina @matteocollina]

For years, I’ve been asked whether I’ll ever write a book about @Nodejs.

I’m excited to announce the time has come.

Presenting “The Definitive Guide for Node.js in Enterprise”, a labour of love by myself, @Lucamaraschi, and every single member of the @Platformatic team.

[Tanner Linsley @tannerlinsley]

📣 Time to announce "create-tsrouter-app" or CTA, a drop-in CRA replacement:

- Powered by @Vite_js + @Tan_Stack Router, built by @jherr
- TypeScript/Tailwind/File-Route templates (more coming)
- Pre-ejected with full control over Vite and (soon) upgradeable to TanStack Start

[Aiden Bai @aidenybai]

oh "frontend is easy"?

implement this efficiently (infinite masonry layout w/ scroll support)

[Sébastien Chopin @Atinux]

An open source headless browser rewritten in Zig designed for AI and automation.

• 9x less memory footprint than Chrome
• 11x faster execution than Chrome

Very exciting times.

https://t.co/dc6XrnrvwM

[Vercel @vercel]

Tremor is joining Vercel.

Tremor is a library of React components to build charts and dashboards. All Tremor components, including Blocks, are now free and open source.

https://t.co/pXbEnQf0N4

[Biome @biomejs]

Today we’re happy to share our plans for Biome 2.0 as well as the rest of our roadmap for 2025.

Read all about it here: https://t.co/G6dVgRPq6b

#OpenSource #WebDev #Tooling for #JavaScript #TypeScript #CSS #GraphQL and #GritQL, written in #Rust

[Theo - t3.gg @theo]

Wow they even hit rspack

[Matt Pocock @mattpocockuk]

This is the sweetest LLM example I've put together so far: a working alt text generator.

- 100% TypeScript
- Works either with files in memory or URL's
- Less than 100 lines of code

Full URL with code examples below

[Evan Bacon 🥓 @Baconbrix]

Expo + React Server Components (preview) 🚀

Just open sourced a new movies app demo using Expo Router and React server functions.

Native data fetching, loading states, and error handling have never been this easy!

→ Runs on iOS, Web, and Android

[Theo - t3.gg @theo]

Our new app is now live! Welcome to T3 Chat, the best AI chat app you've ever used.

This demo is all real time. It's actually that fast.

[Matteo Collina @matteocollina]

This is a good, practical benchmark. I hope AWS optimizes their client. It’s incredibly slow, but they would not optimize it (it’s not in their best interest). A long time ago, my team ended up writing a custom AWS client to overcome the problem (not S3 but dynamo, but the problems are generic).

For everyone looking they key points to get similar speeds in Node are:
- use undici.request() instead of stock node http client (or even fetch). You might need to tune the Agent this a bit.
- move the AWS request signing to a separate thread. Those crypto opts on the main thread blocks some I/O to happen.

[Akshay 🚀 @akshay_pachaar]

Don't do RAG!

Imagine loading all the relevant documents into your model before you ask a single question—no more waiting on real-time retrieval or dealing with complicated retrieval pipelines.

This is precisely what CAG does, and it does so remarkably well!

The core idea is to replace real-time document retrieval with preloaded knowledge in the extended context of LLMs. This approach ensures faster, more accurate, and consistent generation by avoiding retrieval errors and latency.

(refer the image below as you read)

Key advantages:

↳ No Latency: All data is preloaded, so there’s no waiting for retrieval.

↳ Fewer Mistakes: Precomputed KV-cache avoids ranking or document selection errors.

↳ Simpler Architecture: No separate retriever—just load the cache and go.

↳ Faster Inference: Once cached, responses come at lightning speed.

↳ Higher Accuracy: The model processes a unified, complete context upfront.

But it also has two major limitations:

- Inflexibility to Dynamic Data
- Constrained by Context Length of LLM

Hope you enjoyed reading!

For those who want to dig more, I've shared link to the CAG paper in next tweet!
_____
Find me → @akshay_pachaar ✔️
For more insights & tutorials on AI and Machine Learning.

[Seb ⚛️ ThisWeekInReact.com @sebastienlorber]

🧑‍🎄 Node.js now supports TypeScript by default 🤯 🎄

[Rspack @rspack_dev]

Rspack has reached 1 million monthly downloads. 🎉

[React Native @reactnative]

The New Architecture is here!

React Native 0.76 enables the New Architecture by default:

https://t.co/JhOWla2NWy

[william @wgussml]

🚀 I'm excited to announce the future of prompt engineering: 𝚎𝚕𝚕.

developed from ideas during my time at OpenAI, 𝚎𝚕𝚕 is light, functional lm programming library:

- automatic versioning & tracing
- rich local oss visualization tools
- multimodality native

Read on ⬇️

[IM @ivandevp]

@JoshuaCarter603 Definitely, great idea. Thanks for the suggestion @JoshuaCarter603.

[IM @ivandevp]

Esta semana vi una publicación en X sobre rembg, un paquete de Python que usa modelos de machine learning para eliminar el fondo de una imagen. Así que me animé a probarla y creé un notebook con algunas imágenes de prueba

Abro hilo con links y mis opiniones 👇