Daniel Barrios Mogollón @dbmcoder

Si eres Programador Web y usas IA... ¡Instala esto! Agent Skills de Addy Osmani (de Google) para: ✓ Rendimiento Web ✓ Mejores prácticas ✓ Accesibilidad ✓ SEO Para React, Vue, Angular, Astro o lo que uses. $ npx add-skill addyosmani/web-quality-skills

the four horsemen of the apocalypse

keyboard_arrow_left Previous keyboard_arrow_right Next

for anyone asking where to learn this stuff: • RAG → weaviate.io/blog/introduct… • Agentic RAG → weaviate.io/blog/what-is-a… • AI Agents → anthropic.com/research/build… • Multi-Agent Systems → resources.anthropic.com/building-effec… • LangGraph → docs.langchain.com/oss/python/lan… • LangGraph (code) → github.com/langchain-ai/l… • MCP → modelcontextprotocol.io • Memory Systems → letta.com • Evals → docs.smith.langchain.com • Context Engineering → search "Context Engineering Survey" on arXiv and please skip the "build an ai agent in 10 minutes" videos build something, watch it fail, then figure out why.

aditya @adxtyahq

a week ago

again saying there's never been a better time to work on multi-agent systems. learn rag, orchestration, evals, memory, routing, tool calling, validation loops, fix loops, split learning, context engineering. all of it. getting an llm to answer questions is becoming the easy

45 90 843 366K 955

Toru Miyazaki gave 11 cats with advanced kidney disease an experimental injection. 15 others didn’t get it. A year later, 9 of the 11 treated cats were alive. Only 3 of the 15 untreated cats survived. He just filed for approval, and the drug fixes a defect only cats have. Most cats die from one thing: their kidneys fail. By age 10, 4 in 10 cats already have chronic kidney disease, and by age 15, the rate doubles to 8 in 10. Once diagnosed, a cat has about 2 years left. The reason kidney disease hits cats so hard is a broken protein in their blood. All mammals carry a protein that helps the kidneys clean out waste. In humans and dogs, the protein floats freely and goes to work when the kidneys are in trouble. In cats, it stays stuck to another protein and can’t get loose. So the waste piles up, and the kidneys eventually give out. Miyazaki originally found the protein in 1999, back when he was at the University of Tokyo. He figured out the cat-specific glitch in 2015. The paper he published in the Veterinary Journal in February laid out the trial. The injection is a working version of the missing protein. His company, the Institute for AIM Medicine, filed the approval paperwork with Japan’s Ministry of Agriculture on April 24, 2026. If the review clears, the drug goes on sale in spring 2027. The 30-year lifespan figure in the tweet is Miyazaki’s own projection of what cats could reach without kidney disease. The trial only ran a year, and the average cat today lives 15. Most die from the same disease this injection treats. The research almost died in 2020. After running out of funding during COVID, Miyazaki went public. Cat owners across Japan responded by sending in 300 million yen, around 2 million dollars total. He resigned from the University of Tokyo and worked on the drug full time. The treatment in front of regulators today exists because cat lovers refused to let the research die.

Interesting AF @interesting_aIl

3 weeks ago

An injection that can double a cat’s lifespan to 30 years has been developed Clinical trials have begun, with regulatory approval projected in 2027

keyboard_arrow_left Previous keyboard_arrow_right Next

1K 3K 51K 14.9M 9K

Este repositorio es una joya. Te da todos los pasos e instrucciones para proteger y asegurar tu servidor Linux. Perfecto por si tienes un servidor propio o VPS: github.com/imthenachoman/…

6 algorithms that'll help you load balance traffic: 1 Round robin → Load balancer forwards the request to the servers in a sequential order 2 Weighted round robin → Load balancer forwards the request based on server capacity 3 Least connections → Load balancer forwards the request to the server with the fewest connections 4 IP hash → Load balancer forwards the request based on client's IP address 5 Least response time → Load balancer forwards the request to the fastest responding server 6 Adaptive → Load balancer forwards the request based on server health What else should make this list?

¡Vaya tesoro! Colección de APIs gratuitas de modelos de Inteligencia Artificial. Sin pagos y con límites claros. ✓ +20 modelos disponibles ✓ ChatGPT, DeepSeek, Gemini, Qwen y más ✓ Con requests/minuto y tokens por día → github.com/cheahjs/free-l…

Existe un emulador de Android que se ejecuta dentro de Docker. Te permite lanzar un dispositivo Android completo en tu ordenador con un solo comando. Incluye aceleración por GPU, soporte KVM y ejecución headless. Es gratis y 100% open-source 👇

Today we’re introducing a new Legal Agent in @Microsoft Word, built to support the precision and rigor legal work demands. Every clause matters. Every redline tells a story. That’s why this agent was built to follow the structured workflows lawyers use while keeping them fully in control. Early in my career, I asked for a computer on my desk because I believed technology could change how lawyers work. It did. Today, I believe this next generation of tools will do the same, grounded in trust and responsible use.

I found this curated list of free courses with certifications.

x.com/i/article/2045…

99% of AI Engineer interviews are these 7 concepts repeated: → Prompt Engineering: control the model without touching weights → RAG: inject fresh knowledge, kill hallucinations → Fine-Tuning: own the model's behavior end-to-end → Vector Embeddings: semantic search at machine speed → Chain-of-Thought: force step-by-step reasoning, unlock hard problems → Eval Metrics: BLEU, ROUGE, LLM-as-judge. Measure or you're guessing. → MLOps: CI/CD, drift detection, production pipelines. Ship it or it doesn't exist. Master these 7. Pass 90% of AI interviews.

New course: Spec-Driven Development with Coding Agents, built in partnership with @jetbrains, and taught by @paulweveritt. Vibe coding is fast, but often produces code that doesn't match what you asked for. This short course teaches you spec-driven development: write a detailed spec defining what to build, and work with your coding agent to implement it. Many of the best developers already build this way. A spec lets you control large code changes with a few words, preserve context across agent sessions, and stay in control as your project grows in complexity. Skills you'll gain: - Write a detailed specification to define your mission, tech stack, and roadmap, giving your agent the context it needs from the start - Plan, implement, and validate features in iterative loops using a spec as your agent's guide - Apply the same repeatable workflow to both new and legacy codebases - Package your workflow into a portable agent skill that works across agents and IDEs Join and write specs that keep your coding agent on track! deeplearning.ai/short-courses/…

La verdad es que lo siento por la frase pero sigo pensando que destruye el rendimiento de la página sin necesidad alguna y que lo mejor es que la gente despierte sobre eso. 😂 Pero estoy de acuerdo que todo son trade-offs. Sólo que ahora mismo los de Google Fonts son demasiados. Eso sí, también se pueden bajar las fuentes de Google Fonts y hospedarlas y ya está, sólo que Font Source es mucho más cómodo para hacerlo. Aquí tienes una fuente con todas las explicaciones que necesitas más detalladas: martinkoss.com/stop-using-goo… Por otro lado, creo que es importante hablar de esto, ya que mucha gente tiene la falsa creencia que la MEJOR manera de usar fuentes en tu web es Google Fonts. Todavía hay gente que cree que se cachean entre usuarios y es algo que dejo de hacerse por seguridad desde hace tiempo (wicki.io/posts/2020-11-…) Sin contar a nivel europeo los problemas de GDPR siendo consideradas ilegales si no se pedía permiso del usuario: usecue.com/blog/google-fo… Y es que no me da un tuit para hablar de eso. Creo que es fácil entablar una conversación, preguntar y aportar las ideas en el mismo hilo. Creo que este ha sido una oportunidad perdida para eso. Ah, y Google Fonts sigue siendo útil. Me parece un catálogo genial para encontrar fuentes y para prototipar rápido. :)

@neondatabase @traversymedia Opencode with VSCode or Antigravity.

If I had to start with full stack development in this AI era, I'd learn these concepts: 1. HTML5 2. CSS3 3. JavaScript (ES2026+) 4. TypeScript 5. Responsive Design 6. Accessibility (a11y) 7. DOM Manipulation 8. Browser APIs 9. Fetch API 10. Async/Await 11. Git & GitHub 12. Package Managers (npm, pnpm, yarn) 13. Build Tools (Vite, Webpack) 14. React 15. Next.js 16. Server Components 17. Client vs Server Rendering 18. State Management 19. Context API 20. Redux / Zustand 21. Tailwind CSS 22. UI Libraries 23. Design Systems 24. REST APIs 25. GraphQL 26. API Integration 27. Authentication (JWT, OAuth) 28. Authorization 29. WebSockets 30. Forms & Validation 31. Error Handling 32. Testing (Jest, Playwright) 33. Performance Optimization 34. Code Splitting 35. Lazy Loading 36. SEO Optimization 37. Backend Basics 38. Node.js 39. Express.js 40. API Design 41. Microservices 42. Serverless Functions 43. Databases (SQL) 44. Databases (NoSQL) 45. ORMs (Prisma, Drizzle) 46. Caching (Redis) 47. File Uploads 48. Cloud Storage 49. Docker 50. CI/CD 51. Deployment (Vercel, AWS) 52. Environment Variables 53. Security Best Practices 54. HTTPS / TLS 55. Rate Limiting 56. Logging & Monitoring 57. Observability 58. Scalability 59. Load Balancing 60. CDN 61. Edge Functions 62. Progressive Web Apps (PWA) 63. Mobile Responsiveness 64. Cross-Browser Compatibility 65. Web Performance Metrics 66. Lighthouse Optimization 67. AI APIs (OpenAI, etc.) 68. Prompt Engineering 69. AI SDK Integration 70. RAG (Retrieval-Augmented Generation) 71. Vector Databases 72. Embeddings 73. AI Chatbots 74. AI Agents 75. Automation Workflows 76. Background Jobs 77. Event-Driven Architecture 78. Webhooks 79. Streaming Responses 80. Real-Time Apps 81. Feature Flags 82. A/B Testing 83. Analytics Integration 84. User Tracking 85. Personalization 86. Payments (Stripe) 87. Subscription Systems 88. SaaS Architecture 89. Multi-Tenancy 90. Role-Based Access Control (RBAC) 91. API Gateways 92. Service Mesh 93. Monorepos 94. Turborepo / Nx 95. Code Quality (ESLint, Prettier) 96. Documentation 97. Technical Writing 98. System Design Basics 99. Debugging Skills 100. Product Thinking (...and more concepts) === 👋 PS - Want a detailed breakdown of each concept? Read right now: → Grab the Modern Full Stack Latest Edition Handbook: codewithdhanian.gumroad.com/l/fzqjct === 💾 Save this for later & RT to help others learn full stack development. 👤 Follow @e_opore + turn on notifications.

@diegomarino @thenetninjauk Acabo de recordar que siempre vuelvo acá para la config inicial. @TheOdinProject theodinproject.com/lessons/founda…

@diegomarino Conciso, directo y completo. @thenetninjauk youtube.com/playlist?list=…

Today marks a historic return to the moon for @NASA! 🚀 We captured this view of NASA’s Space Launch System rocket and Orion spacecraft on the launch pad earlier today. Here we go! (1 & 2) SkySat • Kennedy Space Center, Florida, United States • April 1, 2026

keyboard_arrow_left Previous keyboard_arrow_right Next

En la era de la #InteligenciaArtificial🤖: “¿El PDF está quedando obsoleto?” Hace unos días me encontré con una publicación donde se cuestionaba si el PDF📄estaba en camino hacia el desuso frente a CRD (Compute-Ready Document). 🔸Entre otros aspectos, la publicación se enfocaba en dos puntos: 1⃣. Cuando un modelo procesa un PDF mediante sistemas de recuperación (RAG), el documento se fragmenta sin criterio semántico. Provocando: ☑️Métodos se separan de resultados. ☑️Gráficas de sus leyendas. ☑️Conclusiones sin contexto. Este problema no es solo técnico. Es un tema de #ArquitecturaDeLaInformación y también de integridad científica, ya que, si la IA no distingue estructura ni procedencia, ¿cómo diferencia investigaciones científicas de contenido fraudulento? 2⃣. El Compute-Ready Document (CRD) como reemplazo del PDF. Al ser documentos diseñados, desde su origen, para ser comprendidos por IA, con datos estructurados, metadatos y trazabilidad (DOI, ORCID, revisión por pares). Pero aquí abro el debate👇🏽: 🔹I. Suponiendo que el PDF está rumbo a la obsolescencia, ¿quién debe definir el nuevo formato del conocimiento computable? ¿Las universidades, organismos, editoriales… o las grandes empresas tecnológicas? 🔹II. Cuando una IA genera una respuesta basada en cientos de artículos académicos, ¿cómo se distribuye la autoría, el crédito y la responsabilidad jurídica? ¿Estamos frente a un vacío del #DerechoTIC o ante la necesidad de nuevos estándares de trazabilidad? 🔹III. Si el valor ya no está en “leer el paper” sino en “usar la respuesta generada por IA”, ¿quién captura ese valor? ¿La comunidad científica o las plataformas que intermedian el conocimiento? 🔹IV. En caso de que la IA se convierta en la principal intermediaria del conocimiento científico, ¿quién audita sus sesgos? ¿Quién garantiza que no reordene prioridades científicas según criterios opacos? ☝️Cierro con una pregunta final: ¿Estamos rediseñando la arquitectura del conocimiento pensando en el interés público o en la eficiencia del mercado tecnológico?