vavanessa.dev

i’ve been on x/twitter for almost 20 years, but mostly as a silent reader. posting a bit more now. If you are into tech, AI, programming, feel free to say hi 👋🥰 Mostly here to talk about my open source projects and sass/mobile apps dev journey/marketing let’s connect 🤝

@bt_sofia_ai @steipete I use both warp, terminal.app, ghostty, hyper.. but i’m always bothered and I go back to iterm2

Idk why people use ghostty beyond the cute mascot 1. Leaks 2. Terminal performance is overrated when we just check our terminals every 5 mins when the AI finishes. Memory is at a premium, not speed. 3. Font rendering sucks, everything looks blurry, and the default accepted solution is to bold everything which makes blur worse. I went back to Hyper.

I smell a leak.

wow this is firee 🔥🔥

@pixelprayer Schedule One

What game is this ?

@codi_fyy that is really cool ! trying it right away :D

🚨 BREAKING: Someone just built the exact tool Andrej Karpathy said someone should build. 48 hours after Karpathy posted his LLM Knowledge Bases workflow, this showed up on GitHub. It's called Graphify. One command. Any folder. Full knowledge graph. Point it at any folder. Run /graphify inside Claude Code. Walk away. Here is what comes out the other side: -> A navigable knowledge graph of everything in that folder -> An Obsidian vault with backlinked articles -> A wiki that starts at index. md and maps every concept cluster -> Plain English Q&A over your entire codebase or research folder You can ask it things like: "What calls this function?" "What connects these two concepts?" "What are the most important nodes in this project?" No vector database. No setup. No config files. The token efficiency number is what got me: 71.5x fewer tokens per query compared to reading raw files. That is not a small improvement. That is a completely different paradigm for how AI agents reason over large codebases. What it supports: -> Code in 13 programming languages -> PDFs -> Images via Claude Vision -> Markdown files Install in one line: pip install graphify && graphify install Then type /graphify in Claude Code and point it at anything. Karpathy asked. Someone delivered in 48 hours. That is the pace of 2026. Open Source. Free.

@brockpierson Yeah, from 2022 to 2025, but then I had to work on some ios/apple projects, so I had to leave my dear Arch Linux / Hyprland 😑😭

Be honest, have you ever used Linux for your daily driver operating system?

@melvynx yet, maybe in 5 to 10 years it won't be anymore

AGI is a marketing word.

@Varnika99 @parthfullstack A better tactic is to create markdown prd’s with tasks, subtasks etc, and ask ai to update it frequently and make documentation. So if you hit rate limits, you can continue with any model

@parthfullstack Why so? I thought it's a decent tectic

Why is everyone shifting from claude codex? Should I also consider it?

@sahill_og Minimax 20$ token plan + opencode value is incredible, claude code pro feels like a demo for the same price

I’m ready to invest $20, Which one should I choose? - Claude - Codex - Cursor - Antigravity - GitHub Copilot Which one is more worth it right now?

@PratikSinhatwt Hey 🥰 Github.com/vava-nessa 1500 ⭐️ on my repo free-coding-models

|￣￣￣￣￣￣￣￣￣￣￣￣￣| | Share your GitHub profile. | |＿＿＿＿＿＿＿＿＿＿＿＿＿| \ (•◡•) / \ / —— | | |_ |_

大家好！我的中国朋友们，我是法国人，正在学习中文。我很想和你们聊天😁🇨🇳🇫🇷(with @HelloChineseApp )

I wonder if the upcoming generation is going to increase that phenomenon, what do you think ?

@justbyte_ Anthropic ceo replied to that in a recent interview, from what i remember it was because he wants AI to help / superpower human creativity, not replace it.

Why doesn’t claude have an image or video generation model?

@AvinashSingh_20 github.com/vava-nessa

Share your GitHub profile, I’ll review it and drop feedback!

@Tech_girlll Even in our vibe coding era YDKJS is still a must read

Do people read programming books these days?

Karpathy didn't make a course. He made THE course. 3 hours. Free. Tokenization. Attention. Hallucinations. Tool use. RLHF. DeepSeek. AlphaGo. Every behavior you've ever wondered about in an LLM - where it comes from, why it exists, how it was engineered. The gap between engineers who understand this and engineers who don't isn't technical depth. It's the ability to conceive of entirely different things.

@Samaytwt Convex ou spacetimedb ! Supabase is still very good tho

Be honest As a developer, which database is better in the AI era?

@GithubProjects Github.com/vava-nessa << free coding models npm, code for free with free AI api endpoints

|￣￣￣￣￣￣￣￣￣￣￣￣￣| | Share your GitHub profile. | |＿＿＿＿＿＿＿＿＿＿＿＿＿| \ (•◡•) / \ / —— | | |_ |_

🤯🤯

INSANE. Did you know the #1 GitHub trending repo (68k+ stars) is literally just a CLAUDE.md file 🤯 It uses @karpathy's 4 LLM principles to keep AI in check: → Seek clarity: Always ask before making assumptions. → Stay minimal: Avoid bloat and keep things simple. → Edit surgically: Limit changes strictly to the task at hand. → Define targets: Map out clear goals before starting. I pasted this into my own workspace, and the code outputs are so MUCH TIGHTER now 💥 Free and open-source. repo below ↓

acaban de resolver uno de los mayores problemas de Claude Code 🙇‍♂️ esta herramienta elimina los límites de uso y evita que deje de funcionar cuando alcanzas el límite. más de 47k stars en GitHub, gratis y open-source. te explico cómo instalarla