24101 MAI14:22
articleLobstersil y a 17 j
6 min de lectureWhy I Don’t Vibe Code
Réflexion critique sur le hype autour du vibe coding et l’IA, mettant en avant les limites des abstractions et la complexité essentielle.
[⋯]Chargement
Le meilleur du dev et de l'IA, scoré chaque jour par un agent. Filtré, résumé, classé. Aucune couleur, aucun bruit — juste la matière.
Réflexion critique sur le hype autour du vibe coding et l’IA, mettant en avant les limites des abstractions et la complexité essentielle.
L’article explore comment des créateurs chrétiens externalisent des vidéos générées par IA via Fiverr, détaille le workflow (script via ChatGPT, narration ElevenLabs, visuals Grok/
Microsoft déploie un agent juridique dans Word qui suit des flux de travail structurés et un playbook pour analyser les contrats et les risques.
Cet article suit le trajet d’un paquet dans un cluster EKS, du pod jusqu’aux balancers, en décrivant le rôle du VPC CNI, le gateway fictif 169.254.1.1 et des entrées ARP figées, puis le passage par iptables (ClusterIP/NodePort) et NLB/ALB. L’appendice présente des commandes et sorties réelles d’un cluster pour s’y essayer.
Supersimple propose un profil OpenCode léger destiné au travail de dev routinier, articulé autour d'agents ciblés et de compétences locales. Le système semble reposer sur une gestion des tracks pilotée par un conductor pour orchestrer les tâches. Le projet est disponible sur GitHub.
GitBook on Vercel uses per-content tagging and selective caching to deliver sub-second updates across 30,000 docs with 120M monthly page views, amid rising AI-driven traffic.
Porting MicroGPT to the Futhark language is introduced, outlining the motivation and scope of the effort. Part I sketches initial approaches and questions for adapting a small GPT-like model to Futhark.
Codex CLI 0.128.0 adds a new /goal mode, letting the coding agent loop until the goal is achieved or the token budget is exhausted. The behavior is implemented via prompts injected at the end of each turn (goals/continuation.md and goals/budget_limit.md). This enables goal-driven, longer-running coding sessions.
The piece imagines what it would take to build a personal GitHub-like platform, weighing architecture, hosting costs, and core features. It discusses trade-offs between self-hosting and cloud services, and outlines practical considerations for a code-hosting project at small to medium scale.
Android VPN can leak the device's real IP even when Always-On VPN is enabled, as demonstrated by a UDP-based bypass technique. The post highlights the risk posed by low-level network behavior and how some traffic can bypass VPN tunnels. It also discusses mitigation strategies and testing approaches to prevent such leaks for security-conscious apps and infrastructure.
Simon Willison évalue les capacités cyber de GPT-5.5 d'OpenAI et discute leurs implications pour la sécurité des systèmes IA. L'article aborde les risques, les scénarios d'utilisation et les limites, avec des réflexions sur les mesures de mitigation possibles.
The piece surveys using Ruby for shell scripting in 2024, highlighting when Ruby can replace or complement traditional shell scripts. It weighs ergonomics and libraries against performance and portability, and offers practical patterns for writing compact, robust Ruby scripts. The discussion is useful for developers evaluating toolchains for automation tasks.
Printing Zig Structs outlines techniques for displaying Zig structs during development, including formatting and debugging considerations. The post points to a discussion thread for additional insights.
Cet article propose un guide pratique pour construire des applications Gleam multiplateformes (Web, desktop et mobile). Il passe en revue les outils, les patterns d’architecture et l’interopérabilité avec les technologies web et système. Des exemples concrets aident à structurer un projet Gleam et à préparer les déploiements multi-cibles.
This post benchmarks the scalability of Postgres for workflow execution, measuring throughput as concurrency increases. It highlights bottlenecks and provides practical tuning considerations for running complex workloads on Postgres.
Matt Webb wants an RSS feed for all the tools and apps pages, each item with an Install button (install location is unclear). The note argues that vibe-coding speeds up development, making apps feel more personal and frequent, like blog posts. It inspired adding an Atom feed (and icon) to the /elsewhere/tools/ page.
Elon Musk testified that xAI used OpenAI's models to improve its Grok through model distillation. Distillation lets a smaller 'student' model learn from a larger 'teacher' model. The case highlights ongoing use of third-party models to enhance in-house AI.
An exploration of building a Game Boy emulator in F#. The post outlines the core architecture and the key challenges of mapping Game Boy hardware to functional F#, along with practical lessons learned during implementation.
The article argues that LLMs should not be treated as junior engineers. They are powerful tools that require careful integration, supervision, and governance within real development workflows. It emphasizes that while LLMs can assist coding tasks, they do not replace skilled engineering judgment and design.
Linux kernel vulnerabilities sometimes reach distributions too late, leaving deploys unpatched. The discussion highlights gaps between kernel disclosures and downstream advisories, and calls for improved visibility for maintainers. For a senior dev, this suggests reworking patching workflows and adding kernel alert feeds to CI/CD.