Nazmul

@bridgemindai More like "temporarily increased for another bait and switch round" 🙄

Claude Code rate limits are way better now. I'm at 27% of my session limit and I've been vibe coding all morning with Claude Opus 4.7. 23% of my weekly limit used. A week ago I was hitting limits in 2 hours. Anthropic actually fixed the rate limits.

@sickdotdev One is more than enough

Why LinkedIn has no competitors?

I’ve been wanting to try pi.dev for a while, so this week I finally did something fun with it. I decided to build **PI’s #PHP version… using PI itself** 😂 So I ported the three core packages: - The main agent core - Provider integrations - A minimal but functional shell agent

@ikoichi If I really believed in it, I would work on another project for some time, then come back with a fresh mind.

Honest question, from founder to founder. If you had been working for almost 2 years on a SaaS, that has been stale at $2k MRR for 6 months (and a bit declining too), what would you do? Fix the retention. Explore new distribution channels. Claude was very strict on "sell" 🤔

@tech_can1x @stupidtechtakes @ManjaroLinux For me it has been super stable, has been my primary work setup for about 8 years now.

@nazmulpcc @stupidtechtakes @ManjaroLinux I wouldn’t recommend Manjaro to my Linux literate friends. It’s fallen off so hard since 2021

i might try out linux again again, what distro is supposed to be good?

Or maybe GitHub will adjust? Even with everything going on, I still think GitHub is the best product on the market by a mile. Just spitballing: what if they launched a “GitHub Lite” — a self-hosted version close to Gitea, with strict limits on free usage? It would ease the LLM-generated spam pressure, stay true to the open-source community, and still keep free users happy.

Developers are actually leaving the platform. Probably great opportunity for some new competitor to rise. Or GitLab to do something special to attract more people.

@dhh Yes but do they actually care about such things? I don't think so, would be very happy if proven wrong 🤷‍♂️

Ternus has a unique chance to reset the relationship between developers and Apple: 1) Tear down the App Store tollbooth, 2) reintroduce Boot Camp for M macs so we can run Linux on them, and 3) fix the infuriating 500ms workspace animation delay. Approval rating +50% instantly.

Why are you promoting this as a new project when it appears to be a fork of github.com/muxy-app/muxy? The original is active, and the description + README are identical with no mention of why the fork was created or what was changed. This comes across as suspicious without any explanation.

Ya puedes dejar de usar iTerm2 Este terminal se llama Muxy (rama huxy) y es el terminal nativo para Mac que necesitabas: • Workspaces por proyecto que se guardan solos (tabs + splits + estado) • Tabs verticales con drag & drop y pinning • Splits perfectos + navegación por teclado • +200 temas brutales • Editor de código integrado con highlight • Git integrado (diff y operaciones) • Ridículamente ligero y rápido (nada de Electron) Lo construyeron en SwiftUI + libghostty y hasta tiene app para iOS. Si codeas en Mac… esto te va a cambiar la vida. REPOOO👇

Not sure why @OpenAI still hasn’t released an official Linux app for Codex. I’ve been happily using this modified version on Ubuntu for over a month now, zero issues, works great. With just a bit more effort from the team, it could be perfect. Linux devs would love this! 👀 #OpenAI #Linux #Codex

@Yampeleg Using this for more than a month now, works pretty well. I tried applying the same formula on claude desktop, it opens, regular chat works but claude code rejects with "Unsupported platform: linux-x64" github.com/ilysenko/codex…

Just told GPT-5.5 to port and compile the codex app to Linux. 1. WTF. It actually worked. 2. WTF. That thing is incredible.

Not true at all! You can comfortably start with the $20 plans — they’re more than enough for most small to mid-sized MVPs. Plus, there are several very affordable providers already: @opencode, @nahcrof, @Zai_org, @Kimi_Moonshot and many others. No need to spend big to get off the ground. 🚀

The vibe coding tax nobody talks about: Claude. Cursor. Vercel. AWS. Resend. Stripe fees. you’re $300 down before your first paying user.

@_alejandroao How fast is it? The biggest downside with cheaper models recently is that they’re usually very slow.

DeepSeek V4 has 1M context window & comparable benchmarks. And it's almost FREE right now. In this video: ✅ What DeepSeek V4 ✅ Benchmark breakdown ✅ Use it with Pi Agent Here are some benchmarks 🧵

@zhenbailian @davidtranwd Well if you are already using workers then it probably makes sense. When you start a new project do check @MXroute They provide incredible value for the price they charge.

@nazmulpcc @davidtranwd if i had any real scale i'd bite the bullet but it doesn't matter for now and I already have workers for something else, so it's just pay-per-use

Instead of paying $20 a month for resend, now I can pay $5 for a worker and $0.35 for every 1000 email.

@dev_maks @Yampeleg Secured: Yes, because you don't crack or introduce any third party code here. ToS violation: Probably not because they've been pretty open about where their llms are called from, this is a first party app so shouldn't be any problem.

@nazmulpcc @Yampeleg Secure ? Doesn't violate ToS ?

@zhenbailian @davidtranwd not going to lie, that is a valid reason to not use such a great product. But it's the cheapest & very reliable, so personally I can handle a one time application + copy credentials

@nazmulpcc @davidtranwd because then i need to open the AWS console

@jaga_prasanna @Yampeleg Follow the instructions, or if you have codex/claude cli installed, provide the repo url to your agent and it will do the rest.

@nazmulpcc @Yampeleg this is not working bro can u share the download link

@piyushgarg_dev Probably not cause that would produce better results

Whatsapp API setup is a heachache man. Meta Business APIs are so broken. Are you guys vibe coding? 🥲

@antonioleivag 5.4 mini is great! I usually plan with 5.4 high, then execute it with 5.4-mini and finally one review pass with 5.4 high.

Si estás con Codex con la cuenta de 20$, dos consejos: 1️⃣ GPT-5.3 Codex sigue siendo un modelo increíble para programar (muchos dicen que mejor que 5.4), y es más barato que 5.4. 2️⃣ GPT 5.4-mini es impresionante para su velocidad y precio. Puede hacer muchas cosas. En mi opinión, uno de los modelos más infravalorados. Con estos dos modelos, puedes estirar mucho tu suscripción, y tener muy buenos resultados.