When will Claude Code adopt AGENTS.md only? I get that CLAUDE.md = "immutable system rules" for better adherence, but maintaining two context files is unnecessary I think. Yes, you can sync both, but you shouldn't have to. simplicity

Here's my December MCP rotation (the ones I actually use): Documentation → context7, cloudflare-docs Project Management → linear, notion Browser Automation → chrome-devtools Drop yours below 👇

Big move in open source AI: the Agentic AI Foundation (AAIF) is here. Anthropic, Block, and OpenAI founded it under the Linux Foundation. Donations: → MCP (Anthropic) → goose (Block) → AGENTS.md (OpenAI) A neutral home for AI agents🤝 aaif.io

Here's my December LLM rotation: Coding → Grok Code Fast 1, Opus 4.5, Sonnet 4.5, GLM-4.6 Research → Gemini 3.0 Pro Finance → Qwen3 VL 4B (running local) Currently paying for: Claude Pro, Z.AI Coding Plan, Opencode Zen, Perplexity Drop yours below 👇

simple and good feedback to handle possible errors

How would you improve this date picker design? What would you change or add to make it better for both OS? 👇

recommendation of the day, good blogpost @tkdodo.eu tkdodo.eu/blog/designi...

should I try @elysiajs.com ? what would you test to validate its value proposition?

Imagine paying for APIs with no signup, no cards, no API keys, just send value for instant access. That's exactly what the x402 protocol proposes: → HTTP 402 Payment Required becomes a web-native stablecoin pay-per-use layer → Built for humans and AI agents www.x402.org

Is it just me or is PayPal's integration DX terrible? I just implemented PayPal Checkout and having experience with Stripe, the difference is clear, PayPal's DX is frustrating. Integrating a payment gateway shouldn't be complicated and "confusing." Thread 🧵

Micro-details make the difference. Solving again React Native's keyboard common issues with react-native-keyboard-controller, smooth view handling that just feels right for the user. Example: without vs with. See the difference?

Claude Opus 4.5 just dropped: → State Of The Art on coding & agentic tasks → 67% cheaper: $5/$25 per million tokens (was $15/$75) Best model for complex reasoning now accessible to more devs www.anthropic.com/news/claude-...

Probé Nano Banana Pro, el nuevo modelo de Google para generar imágenes Genera texto mas legible, acepta hasta 14 imágenes referencias y genera hasta en resolución 4K. Disponible gratis en la app de Gemini con limitaciones. #ai #gemini #nanobanana

[██████████░░░░░░░░░░] 50% 🎨

In simple terms, here's what the new GPT-5.1 model brings: → Adaptive reasoning: adjusts when to "think" based on complexity → Less technical jargon and better instruction following → 2 versions: Instant (speed/simple tasks) and Thinking (complex tasks)

Protecting route groups in @expo.dev router → way simpler than expected. Few lines. What do you think?

Next 📱

just finished this OpenKlub's multi-step onboarding with @tanstack.com form incredible DX compared to other solutions: cleanly separated UI from logic, type-safe by default, and sync/async it just great huge upgrade. keeping this for future OpenKlub forms 🙂‍↕️

Cool! @github.com

Been diving deep into @turborepo.com lately and TS bundlers are a crucial for compiled package strategies. Found tsdown (tsdown.dev) which tsup now recommends as its replacement. Super smooth transition! What's your go-to TS bundler for libraries?

So I'm leaving Notion for Obsidian and gonna try to make it work for personal stuff + adapt it for team collaboration without paying for sync (for now) Any tips or plugins I should know about?

Anthropic acaba de lanzar Claude Sonnet 4.5 🚀 → Mejor modelo de código del mundo → Checkpoints en Claude Code: guarda progreso y haz rollback → Mantiene foco +30 horas en tareas complejas Mismo precio: $3/$15 por millón de tokens www.anthropic.com/news/claude-...

Descubrí y probé el plan de $3 USD de Z.ia (startup china de IA) que funciona con herramientas de agentic coding como Claude Code, OpenCode, etc. Claude → $20-200/mes Z.ia → $3-15/mes Su modelo GLM-4.5 es open source. Te puede ahorrar hasta un 85% 💸 z.ai/subscribe

GitHub lanzó Spec Kit, para generar código desde specs: → Empieza definiendo QUÉ construir (no el cómo) → La IA planifica la arquitectura → Genera código desde especificaciones Compatible: Claude Code, Copilot, Gemini github.com/github/spec-kit #AI #aidevelopment

Si usas Claude Code, esta novedad te va a interesar: ✍️ Opus Plan Mode combina: → Opus 4.1 para planificar y analizar → Sonnet 4 para ejecutar el código Obtienes la inteligencia de Opus donde realmente importa, manteniendo los costos controlados #AI #Claude #ClaudeCode

OpenAI liberó DOS modelos open-source 👀 - gpt-oss-120b (120B parámetros) - gpt-oss-20b (20B parámetros) - Licencia Apache 2.0 - Uso comercial sin restricciones Estaré probando el de 20B, a ver qué tal la experiencia Pruébalos aquí → gpt-oss.com #AI #GPT

Un mes usando Claude Code: la documentación ya no es solo para humanos 🤖 Es la interfaz entre nuestra intención y la IA. ✅ Bien hecho = acelera comprensión ❌ Mal hecho = IA amplifica el caos Si antes documentar era buena práctica, ahora será supervivencia. ¿Lo han notado?

¿Opiniones y para qué aplicación lo utilizarían?

📹 Nuevo video: Aprende cómo autoalojar tu propio sistema de web analytics con Umami - una alternativa privada y open-source Google Analytics. Tutorial fácil para montar Umami + cómo usar su API. 👉 Mira el video aquí: youtu.be/za_uQy9ozt8 #analytics #selfhost