[{"channel_id":1634145994,"post_id":1836,"date":1782167061000,"forwards":"1","views":"88","text":"If you&#039;re building with AI coding agents, it&#039;s time to learn Git worktrees.<br><br>The biggest advantage of AI agents isn&#039;t that they write code\u2014it&#039;s that they can work in parallel.<br><br>The challenge? Multiple agents shouldn&#039;t be modifying the same working directory.<br><br>That&#039;s where git worktree shines.<br><br>Instead of switching branches or cloning the same repository multiple times, you can create separate working directories that all share the same Git history. Each AI agent gets its own isolated workspace to:<br><br>\u2705 Implement a feature<br>\u2705 Fix a bug<br>\u2705 Update documentation<br>\u2705 Experiment with a refactor<br><br>...all at the same time, without stepping on each other&#039;s changes.<br><br>The result is a workflow that feels much more like collaborating with a team of developers than working with a single coding assistant.<br><br>If you haven&#039;t used Git worktrees before, I highly recommend this excellent tutorial series by Net Ninja:<br><a target=\"_blank\" rel=\"noreferrer nofollow\" href=\"https:\/\/www.youtube.com\/watch?v=Vf_0QpLsFRs&amp;list=PL4cUxeGkcC9iUtQh7Aja3TGfbdd7Z-K0W&amp;index=1\">https:\/\/www.youtube.com\/watch?v=Vf_0QpLsFRs&amp;list=PL4cUxeGkcC9iUtQh7Aja3TGfbdd7Z-K0W&amp;index=1<\/a>","text_length":993,"media":{"root":"\/011\/LAcAAMoWZ2EAAAAAc_VhVqkVMKU","webpage":{"url":"https:\/\/www.youtube.com\/watch?v=Vf_0QpLsFRs&list=PL4cUxeGkcC9iUtQh7Aja3TGfbdd7Z-K0W&index=1","type":"video","title":"Git Worktrees Tutorial #1 - What are Git Worktrees?","site_name":"YouTube","display_url":"youtube.com\/watch?v=Vf_0QpLsFRs&list=PL4cUxeGkcC9iUtQh7Aja3TGfbdd7Z-K0W&index=1","description":"In this series, you\u2019ll learn how to use Git worktrees, a feature that lets you check out multiple branches at the same time in separate working directories.\n\n\ud83c\udf7f\ud83d\udc47 Get the FULL Git & GitHub Masterclass:\nhttps:\/\/netninja.dev\/p\/git-github-masterclass\n\n\ud83c\udf7f\ud83d\udc47 Get the FULL Claude Code Masterclass:\nhttps:\/\/netninja.dev\/p\/claude-code-masterclass\n\n\ud83d\udd25\ud83d\udc47 Get access to ALL Masterclasss & premium courses with Net Ninja Pro:\nhttps:\/\/netninja.dev\/p\/net-ninja-pro\/#prosignup\n\nStarter Project: https:\/\/github.com\/iamshaunjp\/portfolio-worktrees","thumbs":{"m":{"w":320,"h":180,"hash":"1PhPmo2BNVu4e7mumNAm4w&ts=1782234901"},"x":{"w":800,"h":450,"hash":"5v-6H4S-JI861HHc0LFzyA&ts=1782234901"},"y":{"w":1280,"h":720,"hash":"Fv5g8T9wZ0m7zSInL8MIbg&ts=1782234901"},"i":{"bytes":"AXACg|DK7VKEJbaMZA7nFRImWGenerdrJCjsZwGBHGUDc596AIjG4GcL\/wB9ijy5OcKOP9sVf+0Wf\/POP\/wHH+NRBbWa5c4CoR8oCgUAU8EEg9RRStgMQBgDpRQBNZBY8ykZ+U8Z5\/DimNIA5Xy0yGPrRRQAeYvH7lP1\/wAaTeAw+QLg54oooASTG4gdqKKKAP\/Z"}},"embed":{"url":"https:\/\/www.youtube.com\/embed\/Vf_0QpLsFRs","type":"iframe","w":1280,"h":720}}}},{"channel_id":1634145994,"post_id":1835,"date":1782109946000,"views":"27","text":"<b>improve<\/b> is an open-source Agent Skill that audits your codebase and generates implementation plans for AI agents.<br><br>The idea is simple: use a powerful model to analyze the codebase and create a detailed plan, then let cheaper models execute it.<br><br>The workflow:<br>\ud83d\udd0d <b>Recon<\/b> \u2014 learns the project structure, coding conventions, and existing documentation.<br>\ud83d\udd0e <b>Audit<\/b> \u2014 analyzes the codebase across multiple dimensions, including correctness, security, performance, testing, technical debt, dependencies, documentation, and developer experience.<br>\u2705 <b>Vet<\/b> \u2014 validates findings to reduce false positives.<br>\ud83d\udcca <b>Prioritize<\/b> \u2014 ranks findings based on impact, effort, and confidence.<br>\ud83d\udcdd <b>Plan<\/b> \u2014 generates a standalone implementation plan for each finding.<br><br>A few features I found interesting:<br>\u2022 Every finding links to the relevant file and line.<br>\u2022 Plans include verification steps, expected outputs, and clear boundaries.<br>\u2022 Plans are versioned against a specific Git commit to detect drift.<br>\u2022 Can execute plans in isolated Git worktrees and review the results.<br>\u2022 Supports focused audits (security, performance, tests, bugs), quick scans, and deep analysis.<br>\u2022 Can export implementation plans as GitHub issues.<br>\u2022 The skill never modifies your source code during the audit.<br><br>Install:<br><br><code>npx skills add shadcn\/improve<\/code><br><br>Repo: <a target=\"_blank\" rel=\"noreferrer nofollow\" href=\"https:\/\/github.com\/shadcn\/improve\">https:\/\/github.com\/shadcn\/improve<\/a>","text_length":1328,"media":{"root":"\/00c\/KwcAAMoWZ2EAAAAAeq_pIerVx6E","webpage":{"url":"https:\/\/github.com\/shadcn\/improve","type":"photo","title":"GitHub - shadcn\/improve: Use your most capable model to audit your codebase and write plans for cheaper models to execute.","site_name":"GitHub","display_url":"github.com\/shadcn\/improve","description":"Use your most capable model to audit your codebase and write plans for cheaper models to execute. - shadcn\/improve","thumbs":{"m":{"w":320,"h":160,"hash":"IJi5XNWaZJpJuxmfM4WZTw&ts=1782234901"},"x":{"w":800,"h":400,"hash":"cSGnZvrgJslRtHRQrWgbhw&ts=1782234901"},"y":{"w":1200,"h":600,"hash":"QRm_kRs_cbr7OcJgoz80Iw&ts=1782234901"},"i":{"bytes":"AUACg|DXYsDwAfxppZ8cKuf96nkZPWjA\/wAmgCMPJ\/zzH\/fVLcOYoHcdVGafgf5NIyK6FWG5T1BoArW10ZHCsQcjPFFTR20MTBkjVSBjIooYEuKKKKACiiigAxRRRQB\/\/9k="}}}}},{"channel_id":1634145994,"post_id":1834,"date":1782105757000,"forwards":"1","views":"31","text":"Build a local RAG pipeline with Ollama \ud83d\ude80<br><br>New crash course from Python Simplified \u2014 turn any LLM into an expert with fast retrieval over your PDFs.<br><br>Covers setup with LangChain + FAISS, chunking, embeddings, retrieval, and the chat model.<br><br><a target=\"_blank\" rel=\"noreferrer nofollow\" href=\"https:\/\/www.youtube.com\/watch?v=oZYlrooPgvs\">https:\/\/www.youtube.com\/watch?v=oZYlrooPgvs<\/a>","text_length":284,"media":{"root":"\/002\/KgcAAMoWZ2EAAAAAhS9KIzvsUO4","webpage":{"url":"https:\/\/www.youtube.com\/watch?v=oZYlrooPgvs","type":"video","title":"Turn Any LLM Into an Expert \ud83d\udcda RAG Coding Crash Course","site_name":"YouTube","display_url":"youtube.com\/watch?v=oZYlrooPgvs","description":"What if you could take any language model and turn it into an expert by giving it fast access to documents? \ud83e\udd14 Big thanks to HubSpot for sponsoring this video \ud83d\ude4c\n\u2b50 Check out HubSpot's FREE AI Agents Cheat Sheet:\nhttps:\/\/clickhubspot.com\/0acb23\n\nIn this beginner-friendly RAG (Retrieval-Augmented Generation) project, we will build a local AI system that can read custom documents, search through them intelligently, and answer questions using knowledge that was never part of the original model! \ud83d\udcda\u26a1\n\nTo make things fun, we'll use a collection of fictional investigation files inspired by The Lord Of The Rings and The Matrix \u2014 turning an ordinary LLM into a specialized AI detective capable of solving mysteries hidden throughout our documents: analyzing suspicious characters, secret reports, and hidden clues like a real pro! \ud83e\uddd9\u200d\u2642\ufe0f\ud83d\udd76\ufe0f\n\nUnlike Fine Tuning, we won't train the model at all! Instead, we will allow a tiny and not-so-smart model to search through documents very quickly, retrieve the most relevant information, and use it to generate intelligent answers in real time \u2014 as tough we are dealing with a giant super-smart model that requires way more resources and computing power. \ud83d\ude0e\n\nBy the end of this tutorial, you'll have a complete local RAG application running on your own computer using Python, LangChain, Ollama and FAISS \u2014 and you'll finally understand why Fine Tuning and RAG are not the same thing!\n\n\ud83d\udcda What You'll Learn:\n\n* What RAG (Retrieval-Augmented Generation) actually is\n* Loading PDF documents into Python\n* Document chunking and text splitting\n* Creating embeddings from text\n* Storing embeddings with FAISS\n* Retrieving relevant context from a vector database\n* Building a complete RAG pipeline with LangChain\n* Running local LLMs with Ollama\n* Answering questions using custom knowledge\n\n\ud83d\udee0 Tools Used:\n\n* Python\n* Ollama\n* Qwen2.5:1.5B (small enough to run on CPU)\n* BGE-M3\n* LangChain\n* FAISS\n* Jupyter Lab\n\n\ud83d\udea8 Are you an educator?\n\nFeel free to use my video slides and project idea in your classes! \ud83d\ude42\nI've included a high-quality PDF containing all the slides from this video, which you can download directly from YouTube.\nThe complete source code, notebook, and project files are also available on GitHub. You're welcome to use them in your lectures, workshops, and classroom activities.\n\nHappy teaching! \ud83d\udc9a\n\n\ud83d\udd0e Resources & Helpful Tutorials:\n\n\u2b50 Full Code and Starter Files on GitHub: https:\/\/github.com\/MariyaSha\/rag_ollama\n\u2b50 My WSL and Conda Tutorial: https:\/\/youtu.be\/luM5kwH6tjQ\n\u2b50 My Hyperparameter Tuning with Sklearn Tutorial: https:\/\/youtu.be\/-IvNzmrcyUM\n\u2b50 Classroom Slides (PDF): Download directly from YouTube \ud83d\udcc4\n\u2b50 Ollama and FAISS GPU links in the pinned comment \ud83d\udccc\n\n\ud83d\udcbb WSL Environment Setup:\n\nconda create -n rag_env python=3.12\nconda activate rag_env\n\npip install \\\n    langchain \\\n    langchain-community \\\n    langchain-ollama \\\n    langchain-text-splitters \\\n    faiss-cpu \\\n    pypdf \\\n    jupyter\n\n\u2699\ufe0f Module Imports:\n\nfrom langchain_community.document_loaders import PyPDFLoader\nfrom langchain_text_splitters import RecursiveCharacterTextSplitter\nfrom langchain_community.vectorstores import FAISS\nfrom langchain_ollama import OllamaEmbeddings, ChatOllama\nfrom langchain_core.prompts import ChatPromptTemplate\nfrom langchain_core.output_parsers import StrOutputParser\nimport os\n\n\u23f0 Timestamps \u23f0\n00:00 Build a Local RAG Pipeline with Ollama\n01:16 Environment Setup (Ollama, LangChain & FAISS)\n04:29 Load PDF Documents for RAG\n07:24 HubSpot AI Agents Cheat Sheet\n09:05 Chunking Documents for RAG\n11:18 Create Embeddings & Build a FAISS Vector Database\n13:32 Retrieve Relevant Documents with FAISS\n15:26 Connect the Chat Model to Your RAG Pipeline\n20:09 RAG Best Practices \n     20:16 Add Chat History to a RAG Chatbot\n     20:54 Give Your LLM Model Identity (a System Prompt)\n     21:27 Run RAG on a GPU with Ollama & FAISS\n     21:49 Load a Saved FAISS Vector Database\n     22:07 Hyperparameter Tuning for RAG\n22:14 Final Thoughts - Is Elrond Guilty?\n\n#python #pythonprogramming #LLM #AI #RAG #LangChain #Ollama #FAISS #MachineLearning #ArtificialIntelligence\u2026","thumbs":{"m":{"w":320,"h":180,"hash":"OUoc_sqa5luut__YQ3PV9A&ts=1782234901"},"x":{"w":800,"h":450,"hash":"v7aBec3g8JWEbs3ISn2i9A&ts=1782234901"},"y":{"w":1280,"h":720,"hash":"7Hev6HxMBwC7HfOWDKAO8A&ts=1782234901"},"i":{"bytes":"AXACg|CrDYs3Mh2+1KNNdmwHAGe9PhvGlkCRwgn3apGa5DD7i7jjjJxjn+lV1G\/h8yFtPVIhIZN2ewqeazjW3DDcT6H\/ADxQUJDb5wxHbHep5JDLCOCN4GeOlKb7BBdzDkXY5X0oqW8GLhgBjp\/KimDRCpI5BI+lWEnyijYCw6k9+tFFNK4W2LkUitC5xjI6VLFJ5sbJzv4I9OKKKzmrOyN4pODZBqKqYwQORRRRVrYzkf\/Z"}},"embed":{"url":"https:\/\/www.youtube.com\/embed\/oZYlrooPgvs","type":"iframe","w":1280,"h":720}}}},{"channel_id":1634145994,"post_id":1833,"date":1782105567000,"forwards":"2","views":"32","text":"Build 3 production AI agents in Python \ud83d\ude80<br><br>New full course from Tech With Tim using Orkes&#039;s Agentspan framework \u2014 agents you can run in production, not toy demos.<br><br>Covers the architecture, memory + tools, RAG, guardrails, and human-in-the-loop.<br><br><a target=\"_blank\" rel=\"noreferrer nofollow\" href=\"https:\/\/www.youtube.com\/watch?v=zFw19qGAeGo\">https:\/\/www.youtube.com\/watch?v=zFw19qGAeGo<\/a>","text_length":289,"media":{"root":"\/003\/KQcAAMoWZ2EAAAAAvwDoUbFJeyY","webpage":{"url":"https:\/\/www.youtube.com\/watch?v=zFw19qGAeGo","type":"video","title":"Build 3 PRODUCTION AI Agents in Python - Full Course (Agentspan)","site_name":"YouTube","display_url":"youtube.com\/watch?v=zFw19qGAeGo","description":"Get started with Agentspan! https:\/\/agentspan.ai\/?utm_campaign=YouTube-Tim&utm_source=Newsletter&utm_medium=email\n\nMost AI agent tutorials show you demos that fall apart the moment you try to run them in the real world. This full course is different \u2014 we'll build three agents from scratch in Python, and every single one is designed to actually work in production.\n\n\ud83d\ude80 Tools I Use\nGet 10% off with code techwithtim\nOpenclaw setup: https:\/\/www.hostinger.com\/techwithtim\nVPS setup: https:\/\/www.hostinger.com\/techwithtim10 \nWispr Flow (Best AI Dictation): https:\/\/ref.wisprflow.ai\/TechWithTim-jun26\n\n\ud83c\udf9e Video Resources \ud83c\udf9e\nFinished Code (GitHub): https:\/\/github.com\/techwithtim\/Agentspan-Course\nAgentspan GitHub: https:\/\/github.com\/agentspan-ai\/agentspan\/tree\/main\/deployment\/docker-compose\nAgentspan Docs: https:\/\/agentspan.ai\/docs\/why-agentspan\/\n\n\u23f3 Timestamps \u23f3\n00:00:00 | Overview\/Intro\n00:01:00 | Problems with Running Agents\n00:02:04 | What we Need for Production\n00:02:45 | Agentspan Framework\n00:04:50 | Agentspan Architecture\n00:06:04 | Install\/Setup\n00:12:00 | Agent 1 - Memory & Basic Tools\n00:28:36 | Agent 2 - RAG, Guardrails and HIML\n00:57:53 | Agent 3 - Multi-Agent Orchestration\n01:11:04 | Testing Agents\n01:12:52 | Durable Execution\n01:17:18 | Deplyment Overview\n\nHashtags\n#Agentspan #Orkes #Python #AIAgents\n\nUAE Media License Number: 3635141","thumbs":{"m":{"w":320,"h":180,"hash":"qNi5curNTpS0TWM-mvb7uQ&ts=1782234901"},"x":{"w":800,"h":450,"hash":"70Ka4ku_XYFtNMXwh7eAaQ&ts=1782234901"},"y":{"w":1280,"h":720,"hash":"9W7-iS6oPnXjfPQEj7MQXA&ts=1782234901"},"i":{"bytes":"AXACg|DIVSzYFOeMocHFWrJUlZY8gE5zVm4WGC3lUvuZhwCKcpRjp1JV2\/IpxWUkqBkKYPbPNPfTJ0GWKAdOtT27RC2jBjBdeWOByM5\/lU5e3VmDQ56jB2\/40nJDsZk1lNDGZG27R6GirNy6Nav+7UHIwQF6fh9aKL3GUI5GicMjEMO9OlmeVt0jFjRRTstwHLPIF2gjGMfdFKbqUnJKk\/7g\/wAKKKVl2FcY88jgqxGD2CgUUUU7IZ\/\/2Q=="}},"embed":{"url":"https:\/\/www.youtube.com\/embed\/zFw19qGAeGo","type":"iframe","w":1280,"h":720}}}},{"channel_id":1634145994,"post_id":1832,"date":1782047645000,"views":"15","text":"While building my local LLM server with MLX, I came across the oMLX project, which already provides this functionality. Looks super awesome!<br><br><a target=\"_blank\" rel=\"noreferrer nofollow\" href=\"https:\/\/github.com\/jundot\/omlx\">https:\/\/github.com\/jundot\/omlx<\/a>","text_length":172,"media":{"root":"\/012\/KAcAAMoWZ2EAAAAABjR8pgXoYHI","webpage":{"url":"https:\/\/github.com\/jundot\/omlx","type":"photo","title":"GitHub - jundot\/omlx: LLM inference server with continuous batching & SSD caching for Apple Silicon \u2014 managed from the macOS menu bar","site_name":"GitHub","display_url":"github.com\/jundot\/omlx","description":"LLM inference server with continuous batching & SSD caching for Apple Silicon \u2014 managed from the macOS menu bar - jundot\/omlx","thumbs":{"m":{"w":320,"h":160,"hash":"NOZycKvaZ0yDkSGrK9CulQ&ts=1782234901"},"x":{"w":800,"h":400,"hash":"AoOembU-L0-5lxkGTv_5qQ&ts=1782234901"},"y":{"w":1200,"h":600,"hash":"vLafCaxfLaK5Tw4j8Dgh4w&ts=1782234901"},"i":{"bytes":"AUACg|DXZmB4UH8abuk\/uD\/vqnkZPWjA9f1oARSSvzYB+tOpMA\/\/AK6Bgd\/1oAZFIHypI3r1FFScUUAZv2+X+6n5H\/Gj7fLkfKn5H\/GiimIPt8ufup+R\/wAaUX0p42p+RoooewD1u5C4BC\/lRRRUFs\/\/2Q=="}}}}},{"channel_id":1634145994,"post_id":1831,"date":1782034022000,"forwards":"1","views":"38","text":"7 agentic AI loops you can lift straight into your workflow \ud83d\udd01<br><br>Matthew Berman&#039;s new ~10-minute walkthrough of his &quot;Loop Library&quot; \u2014 covering, among others, a sub-50ms page-load loop, an overnight doc sweep, an architecture-satisfaction check, and a logging-coverage loop.<br><br>Library: <a target=\"_blank\" rel=\"noreferrer nofollow\" href=\"https:\/\/signals.forwardfuture.ai\/loop-library\/\">https:\/\/signals.forwardfuture.ai\/loop-library\/<\/a><br><br><a target=\"_blank\" rel=\"noreferrer nofollow\" href=\"https:\/\/www.youtube.com\/watch?v=F4a8aMLb678\">https:\/\/www.youtube.com\/watch?v=F4a8aMLb678<\/a>","text_length":373,"media":{"root":"\/011\/JwcAAMoWZ2EAAAAAc_VhVqkVMKU","webpage":{"url":"https:\/\/signals.forwardfuture.ai\/loop-library","type":"photo","title":"Loop Library: Repeatable AI Agent Workflows | Forward Future","site_name":"Loop Library","display_url":"signals.forwardfuture.ai\/loop-library","description":"Repeatable AI agent prompts with practical checks, proof, and stopping conditions.","author":"Forward Future","thumbs":{"m":{"w":320,"h":168,"hash":"0jflgC4YsAeiWa1H9fhwpA&ts=1782234901"},"x":{"w":800,"h":420,"hash":"NslXzx7nLPyaDJuYqMJsTQ&ts=1782234901"},"y":{"w":1200,"h":630,"hash":"7KETTEbnlrW-NvtY3CSGNA&ts=1782234901"},"i":{"bytes":"AVACg|DUZ9rYCMfp0oWXJwUYe56Uwh\/N3eY4GfugDH8ql3+xoAXg9\/1prxrIpVhkHrzSGZAwXPJp4OaAGxxrGgVBgDtminUUAMIfdncNvpil59vyoooGxw6UfhRRQIB9KKKKAP\/Z"}}}}},{"channel_id":1634145994,"post_id":1830,"date":1782018970000,"views":"47","text":"<a target=\"_blank\" rel=\"noreferrer nofollow\" href=\"https:\/\/www.tiktok.com\/t\/ZP8sWQFYT\">https:\/\/www.tiktok.com\/t\/ZP8sWQFYT<\/a>","text_length":34},{"channel_id":1634145994,"post_id":1829,"date":1782016721000,"views":"31","text":"One feature I found particularly useful is the CLI support. You can start, stop, or cancel transcription from scripts, making it easy to integrate into your own workflows.<br><br>According to the project, Parakeet V3 can run at around 5\u00d7 real-time speed on mid-range hardware, making it fast enough for everyday dictation.<br><br>Install:<br>\u2022 macOS: brew install --cask handy<br>\u2022 Windows: winget install cjpais.Handy<br>\u2022 Or download it from the project&#039;s releases page.<br><br>Repo: <a target=\"_blank\" rel=\"noreferrer nofollow\" href=\"https:\/\/github.com\/cjpais\/handy\">https:\/\/github.com\/cjpais\/handy<\/a>","text_length":490,"media":{"root":"\/00b\/JQcAAMoWZ2EAAAAAgz_tAKfx_ck","webpage":{"url":"https:\/\/github.com\/cjpais\/handy","type":"photo","title":"GitHub - cjpais\/Handy: A free, open source, and extensible speech-to-text application that works completely offline.","site_name":"GitHub","display_url":"github.com\/cjpais\/handy","description":"A free, open source, and extensible speech-to-text application that works completely offline. - cjpais\/Handy","thumbs":{"m":{"w":320,"h":168,"hash":"rR13omjG0w0OLlmlJ9buFQ&ts=1782234901"},"x":{"w":800,"h":420,"hash":"ZqRqoNeyI8csr84Iv9pq6Q&ts=1782234901"},"y":{"w":1200,"h":630,"hash":"5lRwK2Bs7VXn3fGTNiIBHg&ts=1782234901"},"i":{"bytes":"AVACg|DXY4FReYxVsDGPUfSpWICknp9M1HlDnKufqpqRifaV9DSG4GBtH1zS7YuBsYD6GkxFtx5bY\/3TT0HoSRvvXOMc0UqY2jAIHvRSJFIz6\/nRsHqfzNFFAw2D1P5mjaPU\/maKKBABj1\/OiiigZ\/\/Z"}}}}},{"channel_id":1634145994,"post_id":1828,"date":1782016704000,"forwards":"3","views":"113","text":"Handy is a nice open-source tool for offline speech-to-text.\n\nPress a keyboard shortcut, start talking, and your transcript is pasted directly into whatever text field you're using. No cloud, no account, and no telemetry.\n\nWhat I like about it:\n\n\u2705 Runs locally with Whisper or Parakeet\n\u2705 GPU acceleration for Whisper, or CPU-optimized Parakeet V3\n\u2705 Uses Silero VAD to automatically stop when you stop speaking\n\u2705 Global hotkey or push-to-talk\n\u2705 Works on macOS (Intel + Apple Silicon), Windows, and Linux\n\u2705 Supports custom Whisper GGML models\n\u2705 Built with Tauri, Rust, and React\n\u2705 MIT licensed","text_length":591,"media":{"root":"\/006\/JAcAAMoWZ2EAAAAAYzEd0sxmE0E","photo":{"thumbs":{"m":{"w":320,"h":268,"hash":"HC83M5saG0jkrqXmCBJ7YA&ts=1782234901"},"x":{"w":800,"h":671,"hash":"7BRmWHwGR57LZaI4d7vTPQ&ts=1782234901"},"y":{"w":1280,"h":1073,"hash":"get0by_APVEJxHzPknc-WQ&ts=1782234901"},"i":{"bytes":"AhACg|CqPIwC\/m5xzwKhpSOOn6UqEDGRkd+cZoGNoGO+aXJHek3H1oELx2zn60UnXmigCb7QwA4T8qjUjIz0zzgimU9TxigdxCTSbjTt9G\/2NAhvWilJz60UADfeP1ptFFABRRRQA5fvD60UUUAf\/9k="}}}}},{"channel_id":1634145994,"post_id":1827,"date":1781970168000,"views":"59","text":"My newsletter is out!<br><br>This week&#039;s agenda:<br>\ud83d\udd39 Open Source of the Week - DiffusionGemma<br>\ud83d\udd39 New learning resources - Gemma 4 12B MTP local test, Column-Level Data Lineage Engine From Scratch, DuoBench planner\/implementer LLM pair benchmark<br>\ud83d\udd39 Book of the week - Large Language Models: The Hard Parts: Open Source AI Solutions for Common Pitfalls by Th\u00e1rsis Souza, PhD and Jonathan Regenstein<br><br><a target=\"_blank\" rel=\"noreferrer nofollow\" href=\"https:\/\/ramikrispin.substack.com\/p\/diffusiongemma-column-level-data\">https:\/\/ramikrispin.substack.com\/p\/diffusiongemma-column-level-data<\/a>","text_length":458,"media":{"root":"\/013\/IwcAAMoWZ2EAAAAAJC97GOG_yXw","webpage":{"url":"https:\/\/ramikrispin.substack.com\/p\/diffusiongemma-column-level-data","type":"photo","title":"DiffusionGemma, Column-Level Data Lineage Engine, LLMs: The Hard Parts | Issue 93","site_name":"Substack","display_url":"ramikrispin.substack.com\/p\/diffusiongemma-column-level-data","description":"A weekly curated update on data science and engineering topics and resources.","thumbs":{"m":{"w":320,"h":160,"hash":"2yQpd4shzv36m-Bv6PHg1Q&ts=1782234901"},"x":{"w":800,"h":400,"hash":"1WHPlVrZUcGWJtY0_i7BHQ&ts=1782234901"},"y":{"w":1280,"h":640,"hash":"VyT56O0_LoSiTKq4gKtqUw&ts=1782234901"},"w":{"w":1600,"h":800,"hash":"tD5P0IoAGFCU0lLay_PKiA&ts=1782234901"},"i":{"bytes":"AUACg|CqJCOpyB1pkruJMA8elSSmONsZBb0qFnG\/cDk+tNsNepMQ+1c8g9akYjHCgDpmmwMX24xgcmnyuC2dlJMTRFsLMAOR70UZkbIAwB6CindgVZCWIJ605VBXJFFFSUPQlM7eOKk3tnrRRUsY0yNtPNFFFAH\/2Q=="}}}}},{"channel_id":1634145994,"post_id":1826,"date":1781875727000,"forwards":"1","views":"47","text":"ponytail \u2014 a skill that gets AI coding agents to write less code, not more \ud83d\ude80<br><br>New open-source project from Dietrich Gebert. ponytail embeds a veteran senior developer persona into AI coding agents \u2014 &quot;long ponytail, oval glasses, has been at the company longer than the version control&quot; \u2014 so the agent stops over-engineering.<br><br>The README&#039;s example: ask for a date picker, and without ponytail the agent installs flatpickr, writes a wrapper component, adds a stylesheet, and starts a discussion about timezones. With ponytail: <code>&lt;input type=&quot;date&quot;&gt;<\/code>.<br><br>Key features:<br><br>\u2705 Decision ladder the agent follows before writing code \u2014 YAGNI first, then stdlib, then native platform, then installed dep, then one-liner, then the minimum that works<br>\u2705 Safety guardrails preserved \u2014 validation, error handling, security, and accessibility are explicitly never trimmed<br>\u2705 Intensity modes \u2014 <code>lite<\/code>, <code>full<\/code> (default), <code>ultra<\/code>, <code>off<\/code><br>\u2705 Slash commands \u2014 <code>\/ponytail-review<\/code> (current diff), <code>\/ponytail-audit<\/code> (whole repo), <code>\/ponytail-debt<\/code> (track deferred shortcuts), <code>\/ponytail-gain<\/code> (impact scoreboard)<br>\u2705 Works with Claude Code, Codex, GitHub Copilot CLI, Gemini CLI, Cursor, Windsurf, Cline, Aider, Zed, and more<br>\u2705 MIT licensed<br><br>Reported benchmark on a real Claude Code agent editing a FastAPI + React repo (Haiku 4.5, 12 tasks, n=4) vs. the same agent without the skill: ~54% less code on average (peak 94% on the date picker task), ~20% cheaper, ~27% faster, 100% safe.<br><br>Get started in Claude Code:<br><code>\/plugin marketplace add DietrichGebert\/ponytail<\/code><br><code>\/plugin install ponytail@ponytail<\/code><br><br>Repo: <a target=\"_blank\" rel=\"noreferrer nofollow\" href=\"https:\/\/github.com\/DietrichGebert\/ponytail\">https:\/\/github.com\/DietrichGebert\/ponytail<\/a>","text_length":1593,"media":{"root":"\/002\/IgcAAMoWZ2EAAAAAhS9KIzvsUO4","webpage":{"url":"https:\/\/github.com\/DietrichGebert\/ponytail","type":"photo","title":"GitHub - DietrichGebert\/ponytail: Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.","site_name":"GitHub","display_url":"github.com\/DietrichGebert\/ponytail","description":"Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote. - DietrichGebert\/ponytail","thumbs":{"m":{"w":320,"h":160,"hash":"0V47pHx2qF_2wkzc8yPBzw&ts=1782234901"},"x":{"w":800,"h":400,"hash":"zy4xG8wnMCOmPJIK36n-Jw&ts=1782234901"},"y":{"w":1280,"h":640,"hash":"cbSnk7MdLOZdvUTSD_lb0g&ts=1782234901"},"i":{"bytes":"AUACg|DNEEhQOF+U9DmlFtM3RDUYbA7\/AJ0pkY\/xN+dMBtWbK0N1IQWCov3j\/hVatPSr6G3heKY7ctnOM5oAr31l9lYFW3xnv6GirOqXlvPAqQncd2ScdKKQGXRRRTAKKKKACiiigD\/\/2Q=="}}}}},{"channel_id":1634145994,"post_id":1825,"date":1781789808000,"forwards":"1","views":"58","text":"Self-healing CI\/CD pipeline with AI \ud83d\udee0\ufe0f<br><br>New freeCodeCamp tutorial walks through wiring n8n, OpenAI, and GitHub Actions together so pipeline failures are automatically detected, analyzed, and resolved.<br><br>Hands-on: a Node.js \/ Express demo app, a smoke test script, the GitHub Actions pipeline, and the n8n automation side.<br><br><a target=\"_blank\" rel=\"noreferrer nofollow\" href=\"https:\/\/www.youtube.com\/watch?v=vj68el9hRvU\">https:\/\/www.youtube.com\/watch?v=vj68el9hRvU<\/a>","text_length":366,"media":{"root":"\/00e\/IQcAAMoWZ2EAAAAA7MNun_1I8lw","webpage":{"url":"https:\/\/www.youtube.com\/watch?v=vj68el9hRvU","type":"video","title":"Build a Self-Healing CI\/CD Pipeline with AI","site_name":"YouTube","display_url":"youtube.com\/watch?v=vj68el9hRvU","description":"Stop manually debugging your code and learn how to leverage AI to automatically detect, analyze, and resolve CI\/CD pipeline failures. In this course, you will bridge the gap between DevOps and automation by integrating N8N, OpenAI, and GitHub Actions to create a fully autonomous, self-healing workflow.\n\n\u270f\ufe0f Course from @TheTechzeen \nLinkedIn: https:\/\/www.linkedin.com\/in\/farzeen-ali-533479204\/\nWebsite: https:\/\/www.thetechzeen.com\/\n\n\u2764\ufe0f Support for this channel comes from our friends at Scrimba \u2013 the coding platform that's reinvented interactive learning: https:\/\/scrimba.com\/freecodecamp\n\n\u2b50\ufe0f Contents \u2b50\ufe0f\n- 00:00 Introduction to Self-Healing CI\/CD Pipelines\n- 02:00 Recommended Tech Stack Overview\n- 02:29 Workflow Logic and Architecture\n- 03:02 Setting Up the Local Environment (Git & GitHub)\n- 05:09 Building the Node.js and Express Application\n- 08:17 Creating the Smoke Test Script\n- 10:22 Implementing the GitHub Actions Pipeline\n- 19:51 Setting Up the N8N Automation Account\n- 21:56 Repository Setup and Security Secrets\n- 27:31 Testing Failure Detection and AI Analysis\n- 30:48 Fetching Logs and Analyzing Changes\n- 38:54 Generating AI Fixes with OpenAI\n- 46:51 Creating a Git Branch and Pushing Fixes\n- 54:32 Automating Pull Requests\n- 56:51 Email Notifications for Team Updates\n- 58:28 Finalizing Production Deployment\n\n\ud83c\udf89 Thanks to our Champion and Sponsor supporters:\n\ud83d\udc7e @omerhattapoglu1158\n\ud83d\udc7e @goddardtan\n\ud83d\udc7e @akihayashi6629\n\ud83d\udc7e @kikilogsin\n\ud83d\udc7e @anthonycampbell2148\n\ud83d\udc7e @tobymiller7790\n\ud83d\udc7e @rajibdassharma497\n\ud83d\udc7e @CloudVirtualizationEnthusiast\n\ud83d\udc7e @adilsoncarlosvianacarlos\n\ud83d\udc7e @martinmacchia1564\n\ud83d\udc7e @ulisesmoralez4160\n\ud83d\udc7e @_Oscar_\n\ud83d\udc7e @jedi-or-sith2728\n\ud83d\udc7e @justinhual1290\n\n--\n\nLearn to code for free and get a developer job: https:\/\/www.freecodecamp.org\n\nRead hundreds of articles on programming: https:\/\/freecodecamp.org\/news","thumbs":{"m":{"w":320,"h":180,"hash":"Lh41TUG-cPtSudx2Z15aVg&ts=1782234901"},"x":{"w":800,"h":450,"hash":"LE_6oZFr7VrA7knQV9qebg&ts=1782234901"},"y":{"w":1280,"h":720,"hash":"4VXezUvY5ht7hek2H8ycYA&ts=1782234901"},"i":{"bytes":"AXACg|CnAhkYN5QdV4Izjr0qUQxkDbC+Djqee49aiggndN8bbQOnzYpmZlPl7mG09M9K2ZNrEzR7o\/8AUMMLjOO+B\/8AXp4jjLg+SeucZHA9DzUSrIdq+aeeCM1ca2xHxJINo659KzegcyKEyAHegxGeBzminyHfH8xbI6ck0UXK1HJPGYo0kBGzjjuKhnmM0zv0BPT0ooq2tRN6EYJGME8VZjuySFlY7R6UUVLRNiJpSykKAo7470UUUij\/2Q=="}},"embed":{"url":"https:\/\/www.youtube.com\/embed\/vj68el9hRvU","type":"iframe","w":1280,"h":720}}}},{"channel_id":1634145994,"post_id":1824,"date":1781700427000,"views":"25","text":"Hermes agent architecture walkthrough \ud83d\udee0\ufe0f<br><br>New tutorial from Alejandro AO breaks down Hermes, his always-on AI agent \u2014 the agent loop, context construction and compression, gateway integrations with Telegram and Slack, SQLite-backed memory, and scheduled cron jobs.<br><br>Includes a look at the context compression prompt that keeps long conversations workable.<br><br><a target=\"_blank\" rel=\"noreferrer nofollow\" href=\"https:\/\/www.youtube.com\/watch?v=n32qq7Kwzh0\">https:\/\/www.youtube.com\/watch?v=n32qq7Kwzh0<\/a>","text_length":401,"media":{"root":"\/001\/IAcAAMoWZ2EAAAAAVk9ou28Qy0s","webpage":{"url":"https:\/\/www.youtube.com\/watch?v=n32qq7Kwzh0","type":"video","title":"Hermes Architecture EXPLAINED: Memory, Context & Gateways","site_name":"YouTube","display_url":"youtube.com\/watch?v=n32qq7Kwzh0","description":"Hermes is an always-on AI agent with a simple but useful architecture: an agent loop, memory, context construction, gateway integrations, and scheduled cron jobs. In this video, I walk through how Hermes builds context, talks to external services like Telegram and Slack, compresses long conversations, and learns from previous interactions.\n\n---\n\ud83e\udd13 *Topics Covered*\n- Hermes agent architecture\n- Context, memory, and compression\n- Gateways, SQLite, and cron jobs\n\n---\n\u23f1\ufe0f *Timestamps*\n00:00 - Intro\n00:57 - Bird's-eye view of the architecture\n03:49 - Agent loop\n07:31 - Context\n13:28 - Context compression\n18:28 - Context compression prompt\n19:59 - Gateway\n28:00 - Memory\n34:37 - Cron jobs\n\n---\n\ud83d\udd17 Links\n- Written version: https:\/\/alejandro-ao.com\/tutorials\/hermes-agent-architecture\/\n\n---\n\ud83d\udc4b Connect with me\n- My website: https:\/\/alejandro-ao.com\/\n- X (Twitter): https:\/\/x.com\/_alejandroao\n- LinkedIn: https:\/\/www.linkedin.com\/in\/alejandro-ao\/","thumbs":{"m":{"w":320,"h":180,"hash":"_ODVwAOBTo8FfzBrxSZ6Bg&ts=1782234901"},"x":{"w":800,"h":450,"hash":"He6tqUZ3bfWs3V6H0Zwiuw&ts=1782234901"},"y":{"w":1280,"h":720,"hash":"7lVKwt0B4f1E5vkvC-g-Lg&ts=1782234901"},"i":{"bytes":"AXACg|DYJI6DNJ3GD1qleXn2eUKAxOMseflH5VNazefAko4yDkfQ4oAmZtql88AVn757geYkpjB6D0q1NHI6TKXAVh8ue3rUIRNu1cgD0qWyoq421uXeUQySZkBPbqKKhtdp1UkHBGc89eKKOUTZZuo5ZZCqzKqFSrKQe9O0+3NtF5fmCQAkjjFFFUIr6pcHaYFONwG41HC5+xhi3ypwcDmiik0NMqo7ed5kYAIOaKKKpCP\/2Q=="}},"embed":{"url":"https:\/\/www.youtube.com\/embed\/n32qq7Kwzh0","type":"iframe","w":1280,"h":720}}}},{"channel_id":1634145994,"post_id":1823,"date":1781668879000,"forwards":"3","views":"37","text":"GitHub for Beginners \ud83d\ude80<br><br>One of the required skills to start working with AI coding tools such as Claude Code is knowing how to work with Git and GitHub. The agent commits, branches, opens pull requests, and merges on your behalf \u2014 and you still need to read the diff, resolve conflicts, and own the repo.<br><br>If you are looking for a resource to get started, the GitHub team released a free 16-video playlist - GitHub for Beginners.<br><br>The playlist covers topics such as:<br>\u2705 A brief introduction to Git<br>\u2705 Beginner Git commands with examples<br>\u2705 Creating your first repository<br>\u2705 Uploading files and adding code<br>\u2705 Pull requests \u2014 opening and merging<br>\u2705 GitHub Issues, Projects, and Actions<br>\u2705 GitHub Pages, Markdown, and security basics<br>\u2705 Open source contributions<br>\u2705 Git and GitHub in VS Code<br><br>Playlist \ud83d\udcfd\ufe0f: <a target=\"_blank\" rel=\"noreferrer nofollow\" href=\"https:\/\/www.youtube.com\/playlist?list=PL0lo9MOBetEFcp4SCWinBdpml9B2U25-f\">https:\/\/www.youtube.com\/playlist?list=PL0lo9MOBetEFcp4SCWinBdpml9B2U25-f<\/a>","text_length":869,"media":{"root":"\/013\/HwcAAMoWZ2EAAAAAJC97GOG_yXw","photo":{"thumbs":{"m":{"w":320,"h":208,"hash":"20MEkbWz23cDIAfSl9ZVnQ&ts=1782234901"},"x":{"w":800,"h":519,"hash":"I1y-iurbzavnLt5b-CzKjQ&ts=1782234901"},"y":{"w":1280,"h":831,"hash":"mooUEjOqvPfnIUXcNCGypQ&ts=1782234901"},"i":{"bytes":"AaACg|CkqrjmJyeo54pbW1a4ZxlRtH8Wf6Uhki8v7jb8dd3H8qsadzJLwuMDr+NU0BSmTy5GTK5HoKjP1FWLvi4b5vy+lQE+5qQG0UUUAOzV\/S\/vyfMB06n61n1d0z77\/T\/Gm2BFeFvtDZLE8dfpUB\/GrOpf8f0g+n8hVSkAUUUUAf\/Z"}}}}},{"channel_id":1634145994,"post_id":1822,"date":1781613287000,"views":"84","text":"Humans can hallucinate, LLMs cannot.<br><br>The term hallucination, in the context of LLMs&#039; output, is fundamentally incorrect. <br><br>LLMs are probabilistic models. When context is missing, they simply fill the gap by making assumptions based on likelihood. Sometimes those assumptions are correct, but often they are wrong. <br><br>This is why, when working with AI agents, context is critical for the agent&#039;s success. <br><br>This is part two of the Foundations of SQL AI Agents series, which focuses on context.<br><br><a target=\"_blank\" rel=\"noreferrer nofollow\" href=\"https:\/\/ramikrispin.substack.com\/p\/context-in-sql-ai-agents-the-three\">https:\/\/ramikrispin.substack.com\/p\/context-in-sql-ai-agents-the-three<\/a>","text_length":563,"media":{"root":"\/003\/HgcAAMoWZ2EAAAAAvwDoUbFJeyY","webpage":{"url":"https:\/\/ramikrispin.substack.com\/p\/context-in-sql-ai-agents-the-three","type":"photo","title":"Context in SQL AI Agents: The Three Layers Behind Reliable Answers","site_name":"Substack","display_url":"ramikrispin.substack.com\/p\/context-in-sql-ai-agents-the-three","description":"Part 2 of the Foundations of SQL AI Agents series","thumbs":{"m":{"w":320,"h":160,"hash":"8CKFtxCZHbyOcg3LrVxesA&ts=1782234901"},"x":{"w":800,"h":400,"hash":"VcFUUqmzoLH8sWpBwukTsg&ts=1782234901"},"y":{"w":1280,"h":640,"hash":"swH7LaN0-cLo8ogEldPF3Q&ts=1782234901"},"w":{"w":1600,"h":800,"hash":"nGczr-81Be7UjUnzmmm45g&ts=1782234901"},"i":{"bytes":"AUACg|Cz85Pb8V\/+vTwrZ+8v5Um9KN6U7gP5C+ppAzEDj8jTS67SP50wOABwvSkBNk0UzzB7UUAUfMal3t60UUAIx3KVbkEUzyYwPuiiigBvlR4J2CiiigD\/2Q=="}}}}}]