The Deep Dive on #AI Agent protocols that @emmanuelbernard.com @meteatamel.bsky.social and I delivered is already online! Big focus on #A2A and #MCP! www.youtube.com/watch?v=DiZs...

We’ve been hinting at it for weeks… now it’s official 🎉 Gemini CLI extensions are here! gemini extensions install ... Super excited to see what the community makes!! blog.google/technology/d...

Super explication , avec une analogie « book publishing », sur AT Protocol. Malgré ce que fait BlackSky, les personnes qui sont bannis de Bluesky sont quand même « grillés ». À moins que tout le monde migre sur BlackSky (improbable) 1/2

Wild but clever idea: to allow backwards compatibility in an evolving data format, embed WASM decoders along with the data to ensure compatibility with a well-know base format (here Arrow). That way older readers for that format can read new encodings, at the price of a small performance hit.

Ça serait intéressant d’avoir un debunk de ce qui est dit sur Cigéo par Pauline Boyer (GreenPeace) dans La Bande Originale avec Nagui aujourd’hui. Notamment sur la pollution qualifiée d’inéluctable des nappes phréatiques. Cc @tristankamin.bsky.social et @lereveilleur.bsky.social

Comment mettre en place une infrastructure de stockage de 30 Po à San Francisco pour du pré‑entraînement de modèles vidéo à moindre coût. (L) "Instead of paying the $12 million / yr … to bring that cost down ~40× to $354k per year, including depreciation." 👉 https://si.inc/posts/the-heap/

I just learnt about the `--resolve` curl option. It binds DNSname+Port pairs with IPs curl a.example.com:9443 --resolve a.example.com:9443:172.18.0.7 Perfect for HTTPS when using multiple containers on the same machine No need for '--insecure' anymore

How to build incredible documentation by @leerob.com leerob.com/docs 1⃣ Fast 2⃣ Readable 3⃣ Helpful 4⃣ AI-native 5⃣ Agent-ready 6⃣ Polished 7⃣ Localized 8⃣ Responsive 9⃣ Accessible 🔟 Universal

Wow, I had no knowledge about that (info leak via Certificate Transparency logs)! It’s not an issue for any public subdomains but really is for ones with private info in it (customer name for instance). A blend of wildcard (for sensitive subdomains) + classic for others could be a compromise

Want a faster Gemini CLI that's smarter with your quota? We're rolling out an experimental Intelligent Model Routing feature. It saves you time and quota by sending simple queries to Flash & complex ones to Pro. How to try it: 🧵 #GeminiCLI #OSS #AI #Gemini

Gemini CLI weekly update for v0.5.0: 🎉 FastMCP integration, a "Three-Day Drop" incoming [█▒▒▒▒▒▒▒▒▒] 0%, positional prompts, custom witty loading messages, and major Edit Tool improvements. Deets: github.com/google-gemin... #GeminiCLI #Gemini #AI #OSS #FastMCP

With `cagent` (github.com/docker/cagent) and the @docker.com MCP Gateway, I can reuse my MCP servers to create a snippets generator and buddy expert in #Golang and #Rustlang in a flash youtu.be/YmQFATFuWAQ

oh my god this is just crazy folks github.com/huggingface/...

simdjson est un projet écrit en C++et dis licence Apache 2 permettant de parser de très grandes volumétries de JSON à la vitesse de l'éclair en utilisant des instructions simd et des optimisations poussées ⬇️ github.com/simdjson/sim...

Metrics are the quantitative backbone of observability—the numbers that tell us how our systems are performing. This is the third post in our OpenTelemetry naming series, this time we tackle the art of naming the measurements that matter.

JetBrains Koog : « We didn’t just port Python concepts to Kotlin. We reimagined what an AI agent framework could be » github.com/JetBrains/koog

🚀 @kestra.io 1.0 is live! What's new: 🤖 AI Copilot 🧠 AI Agents 🔎 AI-powered docs 🔗 MCP Server 🗺️ New flow dependency graph ✨ No-Code Apps & Tests ✅ Many features out of Beta 🔁 New Git sync ☸️ New Helm charts 🪨 LTS release with 1-year of backports! 👉 Full launch post: go.kestra.io/release-1-0-x

OpenTelemetry support in Gemini CLI Great move to precisely track Token usage and provide data for FinOps for AI analysis bsky.app/profile/ntay...

You know that one system? The one running in a corner, doing a critical job? It’s reliable, but it’s also a black box. No one touches it in case it breaks.🫣 But what if there's a low-risk way to see inside? Our latest blog post shows how to instrument legacy apps to gain visibility. buff.ly/lviDk7m

What are OKLCH #colors? jakub.kr/components/o... #frontend

WhisperLiveKit : un projet de speech to text basé sur le modèle Whisper permettent de faire de la transcription en temps réel, directement dans le navigateur avec une interface web simple ⬇️ github.com/QuentinFuxa/...

Les prebuilds avec les Dev Containers c'est quoi et comment ça marche ? 🤔 J'ai plongé dedans pour continuer à fabriquer ma stack de dev et c'est ici que ça se passe : philippart-s.github.io/blog/code/do... 📖 👋

For my series of articles on #ADK for #Java, my colleague Romin Irani used the new "Video Overview" feature of #NotebookLM to generate this great overview of my articles! It's impressive! www.youtube.com/watch?v=HqNS... The recap of the series can be found here: glaforge.dev/posts/2025/0...

The is diabolical... a Python object that hallucinates method implementations on demand any time you call them, using my LLM Python library github.com/awwaiid/grem...

Et si spécialiser un LLM ce n'était si compliqué que ça 🤔 ? C'est ce que je vous propose de découvrir avec mon dernier blog post : 📖 blog.ovhcloud.com/fine-tune-an... Vous verrez qu'avec les Machine Learning Services d'OVHcloud et Axolotl AI c'est un jeu d'enfants 🤖!

Westinghouse und Google Cloud bündeln 75 Jahre Nuklearerfahrung in einem KI-Modell, das Baupläne, Vorschriften und Betriebsdaten verknüpft. Ziel: Bauzeiten verkürzen, Risiken minimieren, Projekte wiederholbar umsetzen, Kosten senken. world-nuclear-news.org/articles/wes...

Not every developer task requires a general-purpose LLM. We’re betting on specialized focal LLMs – smaller, faster, and focused. Join @jetbrains and @huggingface for a livestream on how focal models like Mellum will shape the industry. 📅 July 29, 6 pm CET 👉 Save your spot: jb.gg/uvl8n5

A nice video from @smartereveryday.bsky.social about safety in nuclear power plant (for all public) www.youtube.com/watch?v=cRaK...

Cómo es por dentro y funciona una torre de refrigeración. Créditos: todologia (Instagram): www.instagram.com/todologiaa?i...

Carrot Cache: High-Performance, SSD-Friendly #Caching #Library for #Java

Techno de stockage intéressante ! Espérons que le test en CA confirme ça « à l’échelle » ! L’idée : placer des sphères en béton dans l’océan et les vider/remplir en actionnant une turbine/alternateur pour produire/stocker de ⚡️ Comme une STEP sous-marine. www.iee.fraunhofer.de/en/presse-in...

"If AI is so good at coding... where are the open source contributions?" I think that's a great question! pivot-to-ai.com/2025/05/13/i...

1. LLM-generated code tries to run code from online software packages. Which is normal but 2. The packages don’t exist. Which would normally cause an error but 3. Nefarious people have made malware under the package names that LLMs make up most often. So 4. Now the LLM code points to malware.

I made this free and open source dev tool: gitingest.com Replace "hub" with "ingest" in any github URL to get a text digest that you can feed into any LLM

Tired of replacing "hub" with "ingest" in Github URLs manually? Me too! Well it's over now, @lcandy.bsky.social made a neat browser extension for gitingest! Thank you so much Check out the repo: github.com/lcandy2/giti... Get it for Chrome: chromewebstore.google.com/detail/git-i...

TIL that the #SQLite JDBC Driver, that ships actual binaries of SQLite for different operation systems, much like the #DuckDB one does, has built-in support for @graalvm.org #native image: github.com/xerial/sqlit... via JNI. Pretty cool. #Java #Graal #Databases

S3 has finally added `put-if-match` for compare-and-swap! Expect AWS turbopuffer regions very soon aws.amazon.com/about-aws/wh...

Great read on the performance gains brought by the use of Apache Arrow for data transfer. They may not be where you expect: bandwidth for sure, but primarily on CPU+memory usage to serialize and deserialize data since this is a zero-copy, zero-serialization format. arrow.apache.org/blog/2025/01...

Avec 1 675 MW net de puissance, l'EPR Flamanville 3 devient aujourd'hui le réacteur nucléaire le plus puissant du monde raccordé au réseau. Mais l'EPR est bien plus qu'un nouveau réacteur, pour au moins trois raisons:

‘uv run’ everything with Python! « uv can work within a shebang line and can read a special comment at the top of a Python file that tells uv which Python version to run a script with and which dependencies it needs » treyhunner.com/2024/12/lazy...

Wow. You have to try this. Open aistudio.google.com/live Turn on your microphone and share your screen. And have a real time conversations with the AI about what you are currently working on.

Amazon S3 just grew "append"! It's only available for the more expensive, lower latency S3 Express One Zone bucket class but you can now append data to an object up to 10,000 times - previously you could only atomically replace a whole object with an updated version simonwillison.net/2024/Nov/22/...

Anybody know if it's possible to create a Bluesky List from a Starter pack ?

Le whataboutisme ça me fatigue. Oui il y a pas de solution parfaite (tu veux faire une transition sans acier, ni béton ni travaux ? Bon courage). Non ça ne sert à rien de regarder un élément du système sans regarder le reste (mêmes discussions sans fin avec LCOE et Lazard)

The breton company who grows, grows, grows ! spacenews.com/unseenlabs-e...

Image generation in #Java, with the #Imagen model, and #LangChain4j 🦜☕️ glaforge.dev/posts/2024/0... #GenerativeAI

The power of conventional #git commits by Julien Ponce julien.ponge.org/blog/the-pow...