24 KiB
24 KiB
Amr Inbox
- Michael on X: "Garry is kinda correct here, but is oversimplifying memory. Harrison (the author of the original article) makes a very good point but also makes memory sound easier than it is. (before reading this article, note that I wrote down my thoughts and then passed it through Claude" / X
- Chrys Bader on X: "Why long-term memory for LLMs remains unsolved" / X
- garrytan/gbrain: Garry's Opinionated OpenClaw/Hermes Agent Brain
- karpathy/autoresearch: AI agents running research on single-GPU nanochat training automatically
- Garry Tan on X: "Thin Harness, Fat Skills" / X
- xdevplatform/xmcp: MCP server for the X API
- Winter on X: "Most people don’t realize this: You get 1,500 free daily requests to Gemma 4 31B on @GoogleAIStudio. That’s plenty of free inference (imo). And you can route it into @NousResearch Hermes Agent via Vercel’s AI Gateway: 1. Create an API key on Google AI Studio 2. Add it under" / X
- elebumm/RedditVideoMakerBot: Create Reddit Videos with just✨ one command ✨
- Paweł Huryn on X: "After the llama.cpp fix, we can finally use gemma-4 with Claude Code: Step 1: - Windows: winget install llama.cpp - MacOS: brew install llama.cpp More: https://t.co/2Gq2cdl5hy Step 2: - llama-server -hf ggml-org/gemma-4-E2B-it-GGUF (https://t.co/3kxuATDZL6) OR - llama-server https://t.co/NFIrI1wXtH" / X
- llm-wiki · GitHub
- Kevin Gu on X: "AutoAgent: first open source library for self-optimizing agents" / X
- repowise-dev/claude-code-prompts: Independently authored prompt templates for AI coding agents — system prompts, tool prompts, agent delegation, memory management, and multi-agent coordination. Informed by studying Claude Code.
- Ryan Carson on X: "How to turn your OpenClaw into the world's best assistant" / X
- My self-sovereign / local / private / secure LLM setup, April 2026
- Ahmad on X: "GPU Memory Math for LLMs (2026 Edition)" / X
- Gauri Gupta on X: "auto-harness: Self improving agentic systems with auto-evals (open-sourced !)" / X
- Prompter on X: "CURE procrastination by gamifying your life with AI (prompt included)" / X
- Chrys Bader on X: "Why long-term memory for LLMs remains unsolved" / X
- ValueCell-ai/ClawX: ClawX is a desktop app that provides a graphical interface for OpenClaw AI agents. It turns CLI-based AI orchestration into a desktop experience without using the terminal. China website is https://clawx.com.cn.
- sgoudelis/ground-station: Ground Station is all-in-one satellite monitoring suite
- Token Plan - MiniMax API Docs
- siddsachar/Thoth: Thoth - Personal AI Sovereignty. A local-first AI assistant with integrated tools, a personal knowledge graph, voice, vision, shell, browser automation, scheduled tasks, health tracking, and messaging channels. Run locally via Ollama or add opt-in cloud models. Your data stays on your machine.
- Pacific Pure | Sofas | Product | Design furniture | JORI
- AI Agent Traps by Matija Franklin, Nenad Tomašev, Julian Jacobs, Joel Z. Leibo, Simon Osindero :: SSRN
- Comino Grando RTX PRO 6000 Review: 768GB of VRAM in a Liquid-Cooled 4U Chassis - StorageReview.com
- Elvis on X: "some agentic engineering realization: stop reviewing code. start reviewing contracts. and here's what i mean. the highest leverage work right now is drawing the right boundaries in your system. i've been splitting the system into multiple sub-systems, each with clearly defined https://t.co/Hy0d9WHI1m" / X
- EverMars on X: "Respect to the Man Chasing AI Immortality, Even Though He's Freeloading Off Our Platform" / X
- Seth Howes on X: "I’ve wanted to do this for a decade. But I never did - I refuse to give any company my DNA. It is me. So this week I sequenced my genome entirely at home. Literally on my kitchen table. I never exposed my DNA sequence to the internet. Not at any point. I used a MinION to do https://t.co/qCzXIGxWsy" / X
- All elementary functions from a single operator
- abhigyanpatwari/GitNexus: GitNexus: The Zero-Server Code Intelligence Engine - GitNexus is a client-side knowledge graph creator that runs entirely in your browser. Drop in a GitHub repo or ZIP file, and get an interactive knowledge graph wit a built in Graph RAG Agent. Perfect for code exploration
- An Overnight Stack for Qwen3.6–27B: 85 TPS, 125K Context, Vision — on One RTX 3090 | Medium
- Shiv on X: "A new way to think about composing skills to increase leverage: Skill Graphs 2.0" / X
- Eric ⚡️ Building… on X: "My 4090 went from 26 -> 154 tok/s Qwen 3.6 27B🤯 Same GPU. Same Q4_K_M . No FP8, no extra quant. The unlock: ik_llama.cpp + speculative decoding using Qwen3-1.7B as the draft model. 85% acceptance rate. Full config + benchmarks 👇🏻 https://t.co/NUZec2Y7J0" / X
- Sudo su on X: "for anyone running along, here is the setup. weights from unsloth: https://t.co/b89HJ2JC5O launched with: ./llama-server -m Qwen3.6-27B-Q4_K_M.gguf -ngl 99 -c 262144 -np 1 -fa on –cache-type-k q4_0 –cache-type-v q4_0 same exact flags that held the 3.5 dense baseline at 35.3" / X
- Hasan Toor on X: "Goodbye Claude Code subscription fees. Someone just built a proxy that runs Claude Code completely free… and it's wild. You literally plug in a free NVIDIA API key and point Claude Code at localhost. That's it. It handles everything: - Converts Anthropic API calls to NVIDIA https://t.co/10v4KdymxH" / X
- Graeme on X: "The Local LLM Cheat Sheet for Your 64GB RAM Device We covered 16GB and 32GB already, and 64GB has been the next big request, which I can understand, because this is where things get very interesting. Here's the practical cheat sheet for a 64GB RAM machine for code, math, https://t.co/ohgkjbU0LK" / X
- codejunkie99/agentic-stack: One brain, many harnesses. Portable .agent/ folder (memory + skills + protocols) that plugs into Claude Code, Cursor, Windsurf, OpenCode, OpenClaw, Hermes, or DIY Python — and keeps its knowledge when you switch.
- public-apis/public-apis: A collective list of free APIs
- stanford-iris-lab/meta-harness: Reference code for the Meta-Harness paper.
- thePM001/AEP-agent-element-protocol: AEP (Agent Element Protocol) v2.5 | Deterministic zero-trust total control and governance protocol for AI agents. | Reduce hallucinations to zero through architecure in all constrained domains of application. | LLMs gave you the engines, AEP gives you the control for the thrusters.
- GREG ISENBERG on X: "agreed that the the modern computer probably has to be reinvented 12 "tiny" startup ideas that ride that wave: 1. a "where did I put that" app. you describe what you're looking for in plain english and it searches across every app, folder, email, and slack message you've ever" / X
- iptv-org/awesome-iptv: A curated list of resources related to IPTV
- iptv-org/iptv: Collection of publicly available IPTV channels from all over the world
- iptv-org/epg: Utilities for downloading the EPG (Electronic Program Guide) for thousands of TV channels from hundreds of sources.
- [2604.22709] Thinking Without Words: Efficient Latent Reasoning with Abstract Chain-of-Thought
- 鸟哥 | 蓝鸟会🕊️ on X: "吓到我了！ GitHub上有个工具叫GhostTrack，输入一个手机号，能直接扫出这人在哪些平台注册过账号，还能查IP定位、运营商信息。 克隆代码跑脚本，5分钟搞定，门槛低得离谱。你以为自己在网上隐身呢？ 早被扒得透透的了。信息安全这事，真得重视起来。 🔗 https://t.co/yVGCsmsjyi https://t.co/UyhOY6wHq6" / X
- Where should AI sit in your UI?. Mapping emerging AI UI patterns and how… | by Sharang Sharma | UX Collective
- Shakker on X: "OpenClaw's first output dropped from 1s to 43ms. Plugin bootstrap went from 265ms to 8ms. Provider capability resolution from 49ms to 1.5ms. Config validation from 62ms to 5ms. For most of OpenClaw's history, the plugin system was carrying core, instead of the other way around." / X
- lithos-ai/motus: The open-source agent-serving project
This is the inbox. Everything goes in here when you capture it.