AI Digest

Digest curado

miércoles, 24 de junio de 2026·light·short·7,705 tokens

🔥 TOP — lo que SÍ o SÍ tenés que ver

  • Claude Tag — Anthropic lanzó una nueva feature/producto, probablemente para etiquetar o categorizar conversaciones/agentes. Apenas hay snippet pero es oficial de anthropic.com, revisá ya. link
  • Claude Code Plugins oficiales — Anthropic publicó un directorio oficial de plugins para Claude Code. Esto cambia el juego para tu setup de agentes. link
  • gstack: La setup de Claude Code de Garry Tan — Un repositorio con 23 herramientas opinionadas (CEO, Designer, Eng Manager, etc.) que te dan el mismo workflow que usa Tan para shippear solo como un equipo de 20. Imperdible si usás Claude Code. link

📦 Claude / Anthropic ecosystem

  • Claude Code best practices — Repo curado con tips, agentes, comandos y skills para pasar de "vibe coding" a "agentic engineering". Incluye referencias de Boris Cherny. link
  • ClaudeMeter — App para macOS menubar que trackea tu uso y límites de Claude. Útil si estás activo en la API. link
  • Harness: meta-skill para Claude Code — Un plugin que diseña equipos de agentes especializados a partir de una descripción del dominio. Arquitectura de agentes para proyectos reales. link
  • Hermes Agent (NousResearch) — Agente open-source con loop de aprendizaje: crea skills de la experiencia, busca en conversaciones pasadas, y se construye un modelo del usuario. Corre desde $5/mes. link

🛠️ Dev tools & coding

  • Simon Willison: porting Moebius inpainting a browser con Claude Code — Un case study práctico de cómo usó Claude Code para portar un modelo de image inpainting a WebAssembly. Inspiración directa para tu workflow. link
  • ECC: Agent harness performance optimization — Sistema de skills, instintos, memoria y seguridad para Claude Code, Codex, Opencode, Cursor. Pensado para performance de agentes. link

🏗️ Software engineering

  • "Start with a Monolith" ya no aplica con AI — Post que argumenta que los agentes de AI cambian la ecuación de arquitectura: capaz arrancar con microservicios tiene más sentido ahora. link
  • Prompt Injection as Role Confusion — Paper con blog post companion que redefine inyección de prompts como confusión de roles. Lectura obligatoria si trabajás con agentes. link
  • Why Software Requirements Get Easier in an AI Economy — Reflexión sobre cómo specs se vuelven más flexibles cuando los agentes entienden intención. Directo a tu side project SaaS. link

📚 Vale la pena leer

  • Testing AI-generated code (HN discussion) — Hilo preguntando workflows concretos para testear código generado por AI. Varios compartiendo cómo usan built-in browsers en Codex/Cursor. link
  • BenchPress: predecir score de cualquier LLM en cualquier benchmark — Herramienta de Microsoft para estimar performance sin correr el benchmark completo. Útil para evaluar ruteo de modelos. link

💤 Skippeable pero conviene saber

  • AI stock sell-off — Mercados corrigieron fuerte. Varios titulares (Guardian, AP, Axios). No es impacto directo en lo técnico pero afecta pricing y funding. link
  • Meta's AI reorg backfired — Artículo de Inc. sobre cómo la reorg de Meta en AI salió mal. Contexto institucional, no técnico. link

Artículos fetched (44)

  • Introducing Claude Tag
    anthropic-news· 23-jun

    Jun 23, 2026Product

  • affaan-m/ECC
    github-trending

    The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond. Language: English | Português (Brasil) | 简体中文 | 繁體中文 | 日本語 | 한국어 | Türkçe | Русский | Tiếng Việt | ไทย | Deutsch | Español Warning Official sources only. Install ECC only from verified channels: the GitHub repository github.com/affaan-m/ECC, the npm packages ecc-universal and ecc-agentshield, the GitHub App, the plugin slug ecc@ecc, and the project website ecc.tools. Third-party re-uploads and unofficial mirrors are not maintained or reviewed by the project and may contain malware. 211.9K+ stars | 32.5K+ forks | 230+ contributors | 12+ language ecosystems | Cross-harness agent workflows Language / 语言 / 語言 / Dil / Язык /…

  • jamiepine/voicebox
    github-trending

    The open-source AI voice studio. Clone, dictate, create. Voicebox The open-source AI voice studio. Clone any voice. Generate speech. Dictate into any app. Talk to agents in voices you own. The full voice I/O stack, running locally on your machine. voicebox.sh • Docs • Download • Features • API • Troubleshooting Click the image above to watch the demo video on voicebox.sh What is Voicebox? Voicebox is a local-first AI voice studio — a free and open-source alternative to ElevenLabs and WisprFlow in one app. Clone voices from a few seconds of audio, generate speech in 23 languages across 7 TTS engines, dictate into any text field with a global hotkey, and give any MCP-aware AI agent a voice of your choosing. The two cloud incumbents sit on opposite halves of the voice I/O loop — ElevenLabs o…

  • JCodesMore/ai-website-cloner-template
    github-trending

    Clone any website with one command using AI coding agents AI Website Cloner Template A reusable template for reverse-engineering any website into a clean, modern Next.js codebase using AI coding agents. Recommended: Claude Code with Opus 4.7 for best results — but works with a variety of AI coding agents. Point it at a URL, run /clone-website, and your AI agent will inspect the site, extract design tokens and assets, write component specs, and dispatch parallel builders to reconstruct every section. Demo Click the image above to watch the full demo on YouTube. Quick Start Important: Start by making your own copy with GitHub's Use this template button. Do not clone this template repository directly for your website project, and do not open pull requests here with your generated website. Cr…

  • garrytan/gstack
    github-trending

    Use Garry Tan's exact Claude Code setup: 23 opinionated tools that serve as CEO, Designer, Eng Manager, Release Manager, Doc Engineer, and QA gstack "I don't think I've typed like a line of code probably since December, basically, which is an extremely large change." — Andrej Karpathy, No Priors podcast, March 2026 When I heard Karpathy say this, I wanted to find out how. How does one person ship like a team of twenty? Peter Steinberger built OpenClaw — 247K GitHub stars — essentially solo with AI agents. The revolution is here. A single builder with the right tooling can move faster than a traditional team. I'm Garry Tan, President & CEO of Y Combinator. I've worked with thousands of startups — Coinbase, Instacart, Rippling — when they were one or two people in a garage. Before YC, I was…

  • revfactory/harness
    github-trending

    A meta-skill that designs domain-specific agent teams, defines specialized agents, and generates the skills they use. Harness — The Team-Architecture Factory for Claude Code English | 한국어 | 日本語 Harness is a team-architecture factory for Claude Code. Say "build a harness for this project" (English) or "하네스 구성해줘" (한국어) or "ハーネスを構成して" (日本語), and the plugin turns your domain description into an agent team and the skills they use — picked from six pre-defined team-architecture patterns. Overview Harness leverages Claude Code's agent team system to decompose complex tasks into coordinated teams of specialized agents. Say "build a harness for this project" and it automatically generates agent definitions (.claude/agents/) and skills (.claude/skills/) tailored to your domain. Category — Where Har…

  • shanraisshan/claude-code-best-practice
    github-trending

    from vibe coding to agentic engineering - practice makes claude perfect claude-code-best-practice from vibe coding to agentic engineering - practice makes claude perfect = Agents · = Commands · = Skills Boris Cherny on X (tweet 1 · tweet 2 · tweet 3) Tip Visit the How to Use section to take full advantage of this repo. 🧠 CONCEPTS Feature Location Description Subagents .claude/agents/<name>.md Commands .claude/commands/<name>.md Skills .claude/skills/<name>/SKILL.md Official Skills · Skills for Mono-repos Workflows .claude/commands/weather-orchestrator.md Hooks .claude/hooks/ Guide MCP Servers .claude/settings.json, .mcp.json Plugins distributable packages Marketplaces · Create Marketplaces Settings .claude/settings.json Permissions · Model Config · Output Styles · Sandboxing · Keybinding…

  • anthropics/claude-plugins-official
    github-trending

    Official, Anthropic-managed directory of high quality Claude Code Plugins. Claude Code Plugins Directory A curated directory of high-quality plugins for Claude Code. ⚠️ Important: Make sure you trust a plugin before installing, updating, or using it. Anthropic does not control what MCP servers, files, or other software are included in plugins and cannot verify that they will work as intended or that they won't change. See each plugin's homepage for more information. Structure /plugins - Internal plugins developed and maintained by Anthropic /external_plugins - Third-party plugins from partners and the community Installation Plugins can be installed directly from this marketplace via Claude Code's plugin system. To install, run /plugin install {plugin-name}@claude-plugins-official or brows…

  • NousResearch/hermes-agent
    github-trending

    The agent that grows with you Hermes Agent ☤ Hermes Agent | Hermes Desktop The self-improving AI agent built by Nous Research. It's the only agent with a built-in learning loop — it creates skills from experience, improves them during use, nudges itself to persist knowledge, searches its own past conversations, and builds a deepening model of who you are across sessions. Run it on a $5 VPS, a GPU cluster, or serverless infrastructure that costs nearly nothing when idle. It's not tied to your laptop — talk to it from Telegram while it works on a cloud VM. Use any model you want — Nous Portal, OpenRouter (200+ models), NovitaAI (AI-native cloud for Model API, Agent Sandbox, and GPU Cloud), NVIDIA NIM (Nemotron), Xiaomi MiMo, z.ai/GLM, Kimi/Moonshot, MiniMax, Hugging Face, OpenAI, or your ow…

  • We built telecom infrastructure for AI agents in emerging markets
    hn-ai· 24-jun

    Article URL: https://krosai.com/ Comments URL: https://news.ycombinator.com/item?id=48653504 Points: 2 # Comments: 0

  • "Start with a Monolith" Was Good Advice. AI Is Changing That
    hn-ai· 24-jun

    Article URL: https://medium.com/@pivotfakie/start-with-a-monolith-was-good-advice-ai-is-changing-that-a2181b8e5ede Comments URL: https://news.ycombinator.com/item?id=48655809 Points: 1 # Comments: 0

  • BenchPress: Predict any LLM's score on any benchmark
    hn-ai· 24-jun

    Article URL: https://microsoft.github.io/benchpress/ Comments URL: https://news.ycombinator.com/item?id=48654705 Points: 3 # Comments: 0

  • Ask HN: How do you test AI-generated code?
    hn-ai· 24-jun

    When AI generates code, I first instruct the model to find, fix, and verify any issues. After that, I start the server and test whether it actually works from the user’s perspective. What I’m looking for is a workflow where issues are received, fixed, tested, and deployed—but it seems that current AI agents aren’t very good at performing browser tests from the user’s perspective. I’ve tried using the built-in browsers in Codex and Cursor, but they often only checked whether the page loaded. In the end, I had to instruct them step by step on what to do, and it turned out to be cheaper and faster for me to test it myself. So I’m curious to know how you’ve set up test automation. Are there any services that do this (for individuals, not just enterprises)? If you’re using a harness like Codex…

  • Ask HN: How important is college after AI?
    hn-ai· 24-jun

    Comments URL: https://news.ycombinator.com/item?id=48655137 Points: 2 # Comments: 3

  • Chinese universities are cutting language majors to make way for AI
    hn-ai· 24-jun

    Article URL: https://restofworld.org/2026/chinese-universities-drop-humanities-ai/ Comments URL: https://news.ycombinator.com/item?id=48653897 Points: 10 # Comments: 0

  • GitKraken Unveils Code Flow to Help Teams Navigate the AI Era
    hn-ai· 23-jun

    Article URL: https://sdtimes.com/softwaredev/gitkraken-unveils-code-flow-to-help-teams-navigate-the-ai-era/ Comments URL: https://news.ycombinator.com/item?id=48653306 Points: 3 # Comments: 0

  • Dataland, an intense new AI art museum
    hn-ai· 24-jun

    Article URL: https://www.economist.com/culture/2026/06/23/are-you-having-fun-yet-dataland-an-intense-new-ai-art-museum Comments URL: https://news.ycombinator.com/item?id=48654342 Points: 4 # Comments: 0

  • AI systems out-persuade expert humans
    hn-ai· 23-jun

    Article URL: https://arxiv.org/abs/2606.16475 Comments URL: https://news.ycombinator.com/item?id=48653162 Points: 3 # Comments: 2

  • China Minerals Threatens EU; AI Warfare Dominates Japan, WeChat
    hn-ai· 24-jun

    Article URL: https://asiaai.fyi/wp-login.php?redirect_to=https%3A%2F%2Fasiaai.fyi%2Fwp-admin%2Fpost.php%3Fpost%3D197&reauth=1 Comments URL: https://news.ycombinator.com/item?id=48654155 Points: 3 # Comments: 0

  • Measuring Search Ranking Quality with LLM Judged NDCG
    hn-ai· 24-jun

    Article URL: https://corvi.careers/blog/search-eval-llm-judges-ndcg/ Comments URL: https://news.ycombinator.com/item?id=48653398 Points: 2 # Comments: 0

  • AI Steps Off the Screen
    hn-ai· 24-jun

    Article URL: https://epics.tech/posts/2026-06-23-ai-steps-off-the-screen/ Comments URL: https://news.ycombinator.com/item?id=48655738 Points: 2 # Comments: 0

  • Show HN: Fork.ai – branch any AI answer into a mind map instead of a chat log
    hn-ai· 24-jun

    Article URL: https://forkai.in Comments URL: https://news.ycombinator.com/item?id=48655438 Points: 1 # Comments: 0

  • I built an LLM router that doesn't use an LLM
    hn-ai· 24-jun

    Article URL: https://github.com/itsthelore/wayfinder-router Comments URL: https://news.ycombinator.com/item?id=48655876 Points: 1 # Comments: 2

  • ClaudeMeter – macOS menu bar app to track Claude usage and limits
    hn-ai· 23-jun

    Article URL: https://github.com/yotake/claude-meter Comments URL: https://news.ycombinator.com/item?id=48653089 Points: 3 # Comments: 0

  • Is working in AI training data a waste of time?
    hn-ai· 24-jun

    Article URL: https://joinhandshake.com/ai/opportunities/generalist-bachelor-s2/ Comments URL: https://news.ycombinator.com/item?id=48654916 Points: 1 # Comments: 1

  • Stop asking AI if your startup idea is good
    hn-ai· 24-jun

    Article URL: https://www.idea-launch.io/learn/ai-idea-validator-vs-real-demand-signals Comments URL: https://news.ycombinator.com/item?id=48653414 Points: 3 # Comments: 2

  • 'The Worst It's Ever Been': Why Meta's AI Reorg Backfired Spectacularly
    hn-ai· 24-jun

    Article URL: https://www.inc.com/jessica-stillman/the-worst-its-ever-been-why-metas-massive-ai-reorg-backfired-spectacularly/91363370 Comments URL: https://news.ycombinator.com/item?id=48653507 Points: 28 # Comments: 1

  • FAA awards software and AI contract as part of air traffic control modernization
    hn-ai· 23-jun

    Article URL: https://www.nextgov.com/acquisition/2026/06/faa-awards-software-and-ai-contract-part-air-traffic-control-modernization/414354/ Comments URL: https://news.ycombinator.com/item?id=48653187 Points: 3 # Comments: 0

  • US presses Meta to agree to AI reviews as security fears rise
    hn-ai· 24-jun

    Article URL: https://www.nytimes.com/2026/06/23/business/meta-ai-government-reviews-security.html Comments URL: https://news.ycombinator.com/item?id=48654757 Points: 2 # Comments: 0

  • US AI stock sell-off shakes markets from Wall Street to Asia
    hn-ai· 24-jun

    Article URL: https://www.theguardian.com/business/2026/jun/23/ai-stocks-sell-off-us-markets Comments URL: https://news.ycombinator.com/item?id=48654795 Points: 31 # Comments: 3

  • AI was supposed to make smarter decisions. Iit may be making leaders dangerous
    hn-ai· 24-jun

    Article URL: https://www.youtube.com/watch?v=NTeF1NmmZWA Comments URL: https://news.ycombinator.com/item?id=48654854 Points: 6 # Comments: 2

  • Why Software Requirements Get Easier in an AI Economy
    hn-ai· 24-jun

    Article URL: https://stng.substack.com/p/why-software-requirements-get-easier Comments URL: https://news.ycombinator.com/item?id=48653478 Points: 3 # Comments: 0

  • VoltanaLLM: Energy-Efficient LLM Serving
    hn-ai· 24-jun

    Article URL: https://supercomputing-system-ai-lab.github.io/projects/voltana/ Comments URL: https://news.ycombinator.com/item?id=48655520 Points: 2 # Comments: 0

  • Tech stocks slump as AI bubble fears loom
    hn-ai· 24-jun

    Article URL: https://www.axios.com/2026/06/23/tech-stocks-ai-bubble Comments URL: https://news.ycombinator.com/item?id=48654024 Points: 20 # Comments: 0

  • New AI tools in darktable 5.6
    hn-ai· 24-jun

    Article URL: https://www.darktable.org/2026/06/meet-darktable-5.6-ai-tools/ Comments URL: https://news.ycombinator.com/item?id=48655147 Points: 1 # Comments: 0

  • AI stock slump raises the question are investors taking profits or very nervous
    hn-ai· 23-jun

    Article URL: https://apnews.com/article/tech-stocks-ai-investments-8a0ff4c95d5cae6f65c6e2ba03047058 Comments URL: https://news.ycombinator.com/item?id=48653071 Points: 3 # Comments: 0

  • UN chief urges AI companies to 'come clean' about the pollution they generate
    hn-ai· 24-jun

    Article URL: https://www.fastcompany.com/91563535/un-chief-urges-ai-companies-come-clean-about-pollution-create Comments URL: https://news.ycombinator.com/item?id=48653894 Points: 3 # Comments: 0

  • Get hired faster with data and AI tools to autofill and track
    hn-ai· 24-jun

    Article URL: https://www.froghire.ai/ Comments URL: https://news.ycombinator.com/item?id=48653474 Points: 1 # Comments: 0

  • Red-Teaming after Mythos — Zico Kolter & Matt Fredrikson, Gray Swan
    latentspace· 22-jun

    OpenAI boardmember Zico Kolter and Gray Swan CEO Matt Fredrikson join swyx to explain why AI security is not just “cybersecurity with AI”

  • [AINews] SpaceX is already a $28B/yr Neocloud
    latentspace· 23-jun

    a quiet day lets us reflect on some numbers from Jamin Ball.

  • Prompt Injection as Role Confusion
    simonw· 22-jun

    <p><strong><a href="https://role-confusion.github.io">Prompt Injection as Role Confusion</a></strong></p> First, I absolutely love this:</p> <blockquote> <p>This is a blog-style writeup of the paper.</p> </blockquote> <p>I wish <em>every paper</em> would come with one of these. Academic writing is pretty dry - the impact of a paper can be so much higher if you publish a readable version to accompany the formal one.</p> <p>Charles Ye, Jasmine Cui, and Dylan Hadfield-Menell present some fascinating research into the challenge of having models distinguish their own privileged text (here wrapped in role tags like <code>&lt;system&gt;</code>, <code>&lt;think&gt;</code>, and <code>&lt;assistant&gt;</code>) from untrusted user input wrapped in <code>&lt;user&gt;</code>.</p> <p>The bad news: they…

  • datasette 1.0a35
    simonw· 23-jun

    <p><strong>Release:</strong> <a href="https://github.com/simonw/datasette/releases/tag/1.0a35">datasette 1.0a35</a></p> <p>I'll write more about this one tomorrow, but it's a big release. Three highlights from the release notes:</p> <blockquote> <ul> <li>New "Create table" interface in the database actions menu, backed by the <code>/&lt;database&gt;/-/create</code> <a href="https://docs.datasette.io/en/latest/json_api.html#tablecreateview">JSON API</a>. It can define columns, primary keys, custom column types, <code>NOT NULL</code> constraints, literal defaults, expression defaults and single-column foreign keys. (<a href="https://github.com/simonw/datasette/issues/2787">#2787</a>)</li> <li>New "Alter table" table action and <code>/&lt;database&gt;/&lt;table&gt;/-/alter</code> <a href="ht…

  • OPFS + Pyodide test harness
    simonw· 23-jun

    <p><strong>Tool:</strong> <a href="https://tools.simonwillison.net/opfs-pyodide">OPFS + Pyodide test harness</a></p> <p>I've been pondering if <a href="https://lite.datasette.io/">Datasette Lite</a> - the Python Datasette application run entirely in the browser using Pyodide and WebAssembly - might be able to edit persistent SQLite files stored on the user's computer.</p> <p>That's what <a href="https://developer.mozilla.org/en-US/docs/Web/API/File_System_API/Origin_private_file_system">OFPS</a> (Origin Private File System) is for, so I had Claude Code for web build me this playground UI to try it out in different browsers.</p> <p>Tags: <a href="https://simonwillison.net/tags/browsers">browsers</a>, <a href="https://simonwillison.net/tags/pyodide">pyodide</a>, <a href="https://simonwillis…

  • Porting the Moebius 0.2B image inpainting model to run in the browser with Claude Code
    simonw· 22-jun

    <p>This morning <a href="https://news.ycombinator.com/item?id=48630171">on Hacker News</a> I saw <a href="https://hustvl.github.io/Moebius/">Moebius: 0.2B Lightweight Image Inpainting Framework with 10B-Level Performance</a>, describing a small but effective inpainting model - a model where you can mark regions of an image to remove and the model imagines what should fill the space. The released model <a href="https://github.com/hustvl/Moebius/blob/9310b76e368f5f7a8ecdf06493231af279c9973b/requirements.txt#L1">required PyTorch and NVIDIA CUDA</a>, but since it described itself as 0.2B I decided to try and get it running using WebGPU in a browser. TL;DR: I got it working, and you can try the demo at <a href="https://simonw.github.io/moebius-web/">simonw.github.io/moebius-web/</a>. Read on f…