✍️ I keep having the same conversation with colleagues about using LLMs for code. Finally wrote it down. Planning, context, feedback loops, and why simulation + LLMs is a powerful combo. pierrezemb.fr/posts/llms-for… #llm #rustlang #testing

We hacked the AWS JavaScript SDK, a core library powering the entire @AWScloud ecosystem - including the AWS Console itself 🤯 How did we do it? Just two missing characters was all it took. This is the story of #CodeBreach 🧵👇

My simplified mental model Claude's ToS: 👤 Pro/Max: For you. Local CLI & private scripts acting on your behalf. ⚙️ Usage-based API: For infrastructure. When others trigger the run (CI/CD) or using 3rd-party tools. Am I missing anything? #ClaudeCode

I fully agree. We are not writing code for humans anymore, but for LLMs. Even if, don't get me wrong, Engineers should still own (and so understand) this generated code. Yet, it doesn't need this level of abstraction we used to call "elegant code".

I just discovered BATS (#Bash Automated #Testing System). This is a total game-changer. I can finally stop crossing my fingers and actually add proper tests to all my Bash voodoo! github.com/bats-core/bats…

I noticed Claude was way dumber this morning. For some reasons, it was running under Sonnet 4.5, instead of Opus 4.6. The gap of performances between these two models is insane!

Found a UI library that made me mass mass mass mass mass angry. Angry that this isn't how everything works. Oat: → 6KB CSS + 2.2KB JS → Zero dependencies → No framework required → No build step → Semantic HTML only You write <button>. It looks good. You write <dialog>. It