⌥ eclectic~dev

Keep your code simple and leave the complexity to your fancy chords

@skeptrune she seems happy

"mom, how did we get so poor?" "your father spent our life savings on claude code and shipped nothing"

Hey ChatGPT, help me write a prompt for Claude.

@skane_eng @csaba_kissi @tim_cook

@csaba_kissi @tim_cook There are cheaper Windows laptops with similar features.

Mac just had its best launch week ever for first-time Mac customers. We love seeing the enthusiasm!

haz el esfuerzo por obtener la dopamina de las cosas que creas y no de lo que consumes.

🧲 gifgrep — Grep the GIF. Stick the landing. Search Giphy/Tenor, download GIFs. brew install steipete/tap/gifgrep gifgrep.com Both CLI and TUI. Wanted sth for @openclaw to send me memes. Ended up writing an animated gif renderer for the terminal from scratch. AI can't really 'see' animations so I render a contact sheet with stills.

Otro gran error de Anthropic, OpenCode es muchísimo mejor que ClaudeCode en mis prácticas. OpenCode con Codex es simplemente una joyita.

Anthropic is falling into the same trap as @warpdotdev. They want to add hundreds of features, making the product bloated and not good at any of them. Keep it simple and keep it good

@Bolt__AI is exactly the Mac user experience I was looking for, performant, delicious UI, multi provider, what a great app.

I've had BoltAI on my Mac since the v1 days, but upgrading to v2 last week reminded me why I made the switch. Before Bolt, I was stuck with the ChatGPT desktop app (buggy, missing web features), the Claude app (locked to one provider), and alternatives like MindMac that crashed constantly. There are GitHub threads of people saying it crashes every few messages. What sold me originally was the global shortcut. I hit one key combination and BoltAI appears over whatever I'm working on. No context switching, no tab hunting. But v2 brings even more to the table. The instant workflows are genuinely useful. I set up custom shortcuts for specific tasks, and the inline editing means I can rewrite text or fix code right where I'm typing. Everything is customisable too, from my prompt library to plugins that extend what the AI can do. I can switch between models instantly depending on what I need, using proprietary models or any open source model with the provider of my choice. When I'm working with sensitive client data, I flip to a local model via Ollama and keep everything on my machine. That privacy-first approach matters in some cases. The app isn't perfect. I'd love more detailed cost tracking for API usage (they've added basic usage metrics recently, but more granularity would help). These are nits though. The developer ships regular updates, v2 is noticeably snappier with modern Apple APIs, and it respects your privacy by default. One-time purchase, no subscription. For Mac power users who need AI properly integrated into their workflow, BoltAI v2 is the best option I've found.

Hey @bcherny @claudeai take notes from boltai.com on how to build a very nice/performant desktop experience, Claude desktop is super awful

was messing with the OpenAI base URL in Cursor and caught this accounts/anysphere/models/kimi-k2p5-rl-0317-s515-fast so composer 2 is just Kimi K2.5 with RL at least rename the model ID

Montreal's one of my favourite cities. Super walkable, incredible food, and among the best public transit in NA. Some of the nicest WeWorks you'll find too. I come here every year, sometimes more. x.com/bentlegen/stat…

El "Efecto Cursor": Rápido hoy, inmanejable mañana. Un estudio de la Universidad Carnegie Mellon (publicado en arXiv) ha analizado miles de repositorios que usan Cursor AI. ¿El resultado? Sí, escribes código 3-5 veces más rápido el primer mes, pero a los dos meses la complejidad y los avisos de errores estallan un 41%. Al parecer la IA está llenando GitHub de "spaghetti code" automatizado. Básicamente, estamos pidiendo préstamos de deuda técnica a un interés del 300% que pagaremos en 2027. La velocidad es una ilusión si el código es un laberinto.

The Montreal Renaissance is real. The Habs are ridiculously good. The Expos spark is back. The food scene is booming. The coffee culture is top tier. The bagels...never better. The tech ecosystem is bouncing back. The animation scene is back (just won an Oscar) The history, unmatched. Bullish on this city.🚧

Estaba explorando el indice de uso de Claude de Anthropic, y me parece entre interesante y deprimente que México tenga una de las tasas de uso más bajas de Latam. Chile por ejemplo tiene un indice de uso de mas del doble que México. Esto nos habla mucho de la adopción de la IA como agente de programación. Ya que los principales usos de Claude en México son para temas académicos y de progrmación.

Jugaron contigo, el Pokemon eras tú.

@La_JennyLove @OrbStack @devwithlando

Thanks to @init_sh for the recommendation, since I have switched to @OrbStack with @devwithlando my dev stack flies. ✨

Me cagan todos los clientes de chat AI, Claude, ChatGPT, etc.. todos son SUPER lentos a estas alturas prefiero chatear en opencode hahaha entonces generé una carpeta (proyecto) donde chateo con mi modelo favorito y la neta funciona chingón

“This hour has 22 minutes” (a legendary Canadian sketch comedy show) did a Pitt parody on the Canadian healthcare system. “Robby” using hand sanitiser every 10 seconds is taking me out 😂😂😂