0xNeoArch

PROP TIP Running LLMs locally? Give them web access My setup: - SearXNG: candidate source discovery - Firecrawl: known-URL scraping and crawling - Camofox: browser fallback when JS/interaction gets annoying Search → Extract → Interact Tell your favorite agent to set this up, then wire it into your local models > Watch them suddenly become way more useful You’re welcome

Your AI assistant can now actually browse the internet in real time. This tool gives AI agents like Claude and Cursor access to Twitter, Reddit, YouTube, GitHub and more through a single CLI with zero API fees. If you ever wished your AI could just go look something up on social media itself, this is it. github.com/Panniantong/ag…

NVIDIA just open sourced a security scanner for AI agent plugins and the results are eye opening. Over 26% of agent skills they tested contain real vulnerabilities and 5% show signs of being outright malicious. Run this before installing any third-party AI tool on your machine. github.com/nvidia/skillsp…

GPT-5.6 rumors are everywhere 👀 If OpenAI really ships: • 1.5M token context • Better agent workflows • Lower pricing then the biggest story won't be the model itself. It'll be what autonomous AI agents can build when context windows become effectively "whole-project" sized. June could get interesting. 🚀

4步教你复活 claude fable 5！ 1.下载 fable 5的系统提示词 github.com/elder-plinius/… 2.放到你Claude code 项目文件夹 3. 使用启动命令 claude --dangerously-skip-permissions --system-prompt-file CLAUDE-FABLE-5.md 4.模型切换到 opus 4.8 Max 只需要这4步 你会发现 fable 5 又回来了！

If you have: Hermes Agent Claude Code & Codex Handoffs Obsidian + QMD Memory System Run Agentic Loops Fleet Tailscale Mesh Cron Jobs + Kanban Board Agentic Workflows Congrats you are the top 1% of the AI god stack

Sglang is my #1 inference engine. - most active community - most models supported - best performance for 6000s - good GitHub issues I love you sglang ❤️

If you have the NVMe Go download as many models as you think you might ever want. Now, go on Huggingface. They’re coming for open models next.

One day similar approaches would let us have Fable-level intelligence at home 🚀

Code2LoRA seems an incredibly interesting idea. Qwen2.5-Coder-1.5B is not the most powerful LLM around, but it's enough to validate the concept. Instead of stuffing repository context into the prompt at every query, distill it into a LoRA adapter. One forward pass over the repo snapshot, one adapter, zero extra inference tokens. For evolving codebases, a single layer GRU tracks commit history on top of that snapshot. Each git diff updates the hidden state in <10ms. You get a fresh adapter at every commit without need for a full retraining. Great job Liliana! I bet this will lead to something cool in the near future 🙌

Noooo, when I woke up it was already disabled… We had the best few days 🤯

@AnthropicAI Noo, this was the best model ever!

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Claude models is not affected. We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible. Read our full statement: anthropic.com/news/fable-myt…

El CEO de Anthropic tras darse cuenta de que con su Claude Fable 5 se lanzó este tutorial gratuito que está rompiendo internet donde se puede crear sitios web animados y premiados de forma brutal en solo 12 minutos

@0xSero awesome, will try it! thank you!

@0xNeoArch already available on models tab

From running your local model to droid, pi, opencode, and hermes in 1 click.

@vu3dtu awesome!

@vu3dtu very cool! i want to help with testing. will the Pro version of the ring work, too?

I loved the idea behind the Google Fitbit Air: an LLM wrapped around your health data, daily briefs, and a coach you can ask questions. But there app is really terrible, it's expensive $100 band plus $10/mo, and Google getting a constant stream of your heart rate, sleep, and other private data. Whoop is worse, with a subscription that runs up to $360 a year. So I bought a $7 generic Chinese smart ring from Temu, reverse engineered its BLE protocol, and built an app around it. Introducing PulseLoop: no subscription, open-source iOS app. Your health data stays on your phone, paired with an AI coach that reads your real ring data, draws charts, and remembers context. Free, bring your own API keys. Demo and code below.

We're all sleeping on Step-3.7-Flash. It's phenomenal.

When all these closed labs decide it's time to rug pull everyone, you all are gonna regret not Buying a GPU Owning your compute allows you to be in control, even if partially Not too late yet