56 subscribers
7.26K videos
7.86K links
Download Telegram
This media is not supported in your browser
VIEW IN TELEGRAM
📦 prismml-eng/bonsai-demo

Bonsai Demo

You can now run a massive twenty-seven billion parameter artificial intelligence model directly on a standard smartphone or a normal consumer laptop without running out of memory. The bonsai-demo project achieves this wizardry by compressing large language models down to a tiny one-bit or ternary footprint, packing complex weights so efficiently that they fit into a fraction of the space normally required. It gives you fully localized vision capabilities, deep reasoning, and tool calling with zero external API fees. It is basically the holy grail of offline, private intelligence that actually runs fast on everyday hardware you already own.

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 prettysmartdev/awman

awman

You can now run and coordinate multiple AI code agents simultaneously in container-isolated sandboxes right from your terminal. awman is a terminal interface and API that automates your entire development lifecycle from an open GitHub issue to a merged pull request. It runs different agents like Claude or Codex in parallel tabs, isolating their environments with Docker and Git worktrees so they do not conflict. By defining structured workflows in simple configuration files, you can coordinate step-by-step tasks like planning, coding, and testing. You can even run it in yolo mode to let agents autonomously fix bugs, check CI status, and push pull requests while you watch the progress update.

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 wassermanproductions/wassermans-filmmaker-suite

Wasserman's Filmmaker Suite

Choreograph 3D camera moves and actors on a virtual stage, then export exact motion files that AI video generators can actually understand. Wassermans-filmmaker-suite is a collection of four desktop applications built to bridge the gap between traditional pre-production and modern generative video. Instead of fighting with random text prompts, creators can block out scenes, extract motion data from real footage, and even split audio tracks into separate stems locally. An integrated server system even lets external AI assistants automate editing tasks right inside DaVinci Resolve. It turns chaotic video generation into a precise, controllable filmmaking workflow.

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 nv-tlabs/learning-convex-decomposition

learning-convex-decomposition

Break down complex 3D shapes into simpler, snug-fitting blocks to make physics simulations and collision detection run incredibly fast. The learning-convex-decomposition project introduces a clever feed-forward model that predicts a continuous feature field across a 3D object, which then clusters itself into neat convex parts using a purely geometric, self-supervised process. What makes this so exciting is that it is the first open-world method of its kind, meaning it generalizes beautifully to completely unseen objects, whether they are represented as standard meshes, CAD models, or even Gaussian splats, helping game developers and animators instantly optimize complex 3D scenes.

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 xmanrui/openclaw-bot-review

OpenClaw Bot Review

You can monitor and live-test every single AI agent, model, and communication platform in your setup with a beautiful web dashboard that doesn't even require a database. The openclaw-bot-review project plugs directly into your local configuration file and session data to give you real-time visual tracking of token usage, message statistics, and system alerts. The absolute best part is the pixel-art digital office, where your active AI bots are represented as animated characters that walk around, sit at desks, and interact with the environment. It is the perfect visual command center for anyone running local virtual assistants.

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 yeet-src/sigwire

Sigwire

Sigwire is the live diagnostic tool that finally makes tracing Linux system signals as simple as reading a terminal log. Instead of struggling with intrusive debuggers on a single process, this clever tool hooks directly into the kernel's native tracepoints to stream every signal across your entire machine in real time. It acts like a live patchbay, showing you exactly who sent a signal, who took the hit, whether the target caught it, and even if it broke a waiting system call. It is the ultimate way to catch those silent, timing-dependent bugs instantly without slowing down your system. Check it out to see your operating system's hidden traffic come...

📰 https://news.ycombinator.com/item?id=48898071

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 marcelpanse/youtube-guitar-tab-parser

YouTube Guitar Tab Parser

This clever command line tool automatically extracts sheet music from any YouTube guitar lesson and compiles it into a clean, ready-to-print PDF. Instead of you taking dozens of manual screenshots and cropping out the instructor's hands, the program downloads the video, uses Claude vision to isolate the exact tab region, and strips out the background. It then reads the actual bar numbers on the score to automatically filter out duplicates and transition frames. You get a perfectly sequenced, high-quality document of the song's entire tab layout with zero configuration, leaving you with a clean score to practice your favorite songs.

📰 https://news.ycombinator.com/item?id=48898154

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 poseidon-fan/linux-0.11-rs

linux-0.11-rs

Step back to 1991 and experience the early days of personal computing by booting a complete operating system built entirely with modern safety. By rewriting the original Linux 0.11 kernel from scratch in Rust, this project brings robust type safety to retro systems programming. It boots on emulated hardware and runs a self-hosted userland complete with memory management, a file system, and over eighty core utilities. Best of all, it includes an interactive shell with tab completion and history. It is a brilliant way to explore kernel development without the complexity of modern operating systems. Grab the code, launch the emulator, and start exploring.

📰 https://news.ycombinator.com/item?id=48898134

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 usefulsoftwareco/executor

Executor

Stop copy-pasting API keys and configuring the exact same integrations across Claude, Cursor, and ChatGPT. Executor acts as a central hub for all your AI tools, letting you configure any API, database, or custom script exactly once, set your security policies, and share them across every agent you use. Instead of scattering your credentials everywhere, you get a single secure catalog that connects over the Model Context Protocol. You can run it as a desktop app, a local command-line service, or in the cloud. It is the ultimate upgrade to make your favorite AI assistants actually useful.

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 saganaki22/comfyui-woosh

ComfyUI-Woosh

Generate custom sound effects directly inside your node-based workflows using the power of Sony AI's Woosh model. This clever integration lets you turn text descriptions or even video frames into high-quality audio tracks without wrestling with complicated audio engineering tools. If you have ever wanted to instantly match an explosion, a gentle breeze, or footsteps to your visual creations, this project is a total lifesaver. It is incredibly efficient, offering distilled models that cook up sounds in just four steps while keeping your system memory happy. Grab the model, plug it in, and bring your silent videos to life.

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 sozercan/kaset

Kaset macOS App

Enjoy your favorite music and videos right from your desktop by running YouTube and YouTube Music inside a gorgeous, native macOS app. Kaset completely replaces clunky browser tabs with a beautifully integrated Apple Music style interface, complete with media key support, system equalizer presets, and lock screen controls. Instead of bogging down your machine, it uses lightweight background audio and smart WebKit extensions to block ads and sponsor segments automatically. It is the ultimate way to bring YouTube personalized recommendations, playlists, and synchronized lyrics directly into your Mac ecosystem. Download Kaset today and make your desktop listening experience feel truly native.

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 facebookresearch/meshflow

MeshFlow

Generate highly detailed, artistic 3D meshes in just about one second. This project makes 3D creation incredibly fast by combining a clever shape encoder with a flow-based diffusion system. Normally, generating clean 3D structures from scratch takes massive computational power and tedious waiting. This tool solves that by encoding complex shape structures into simple mathematical coordinates, letting a fast diffusion model generate beautiful meshes from a basic guide shape or an optional reference image. It is perfect for anyone wanting to create high-quality 3D assets instantly without waiting on heavy rendering pipelines. Try it out to speed up your 3D design workflow.

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 extraltodeus/j-wash

j-wash

Reshape an AI model's entire identity and behavior without any traditional training, datasets, or fine-tuning. The j-wash project provides a local interactive studio where you can chat with a model and watch what its internal layers are reading in real-time. By pinpointing specific concepts, you can write simple token-level rules to rewrite how the model thinks, like swapping out its self-identity. The best part is you can bake these edits directly into a standard checkpoint to run anywhere. It is a brilliant way to explore and customize AI weights with immediate, exportable results.

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 tiger-ai-lab/openresearcher

OpenResearcher

An open-source AI agent just outperformed GPT-4 and Claude on deep research tasks by using its own self-built search index instead of relying on expensive external search APIs. This project is called openresearcher, and it is a complete, fully open pipeline designed to handle complex, long-term research journeys. By using a clever local retriever built over an eleven-billion-token corpus, it slashes training and operating costs while letting the AI search the web, browse pages, and reason through heavy tasks entirely on its own. If you want to run massive research agent pipelines locally without blowing your budget on API keys, you should definitely check this out.

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 casberryindia/physics-notebook

Physics Notebook

Physics Notebook is the interactive web library that brings difficult physics concepts to life using browser-native animations instead of heavy frameworks. Users can explore complex topics like orbital mechanics, wave interference, and quantum double-slit experiments using smooth, real-time canvas simulations. What makes this project incredibly cool is its zero-dependency architecture, relying on pure vanilla JavaScript for ultra-fast performance. Every visualization includes adjustable sliders that let you manipulate physical variables, alongside beautifully rendered mathematical equations and quick-access AI prompts to immediately query concepts, making it a highly responsive, visual playground for anyone curious about how the universe actually works.

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 tej-sharma/selfloop

Selfloop

Your app testing can now run on complete autopilot. Selfloop is an intelligent visual AI engine that drives your iOS simulator or web browser just like a human tester would, tapping buttons, filling out forms, and exploring every corner of your application. Instead of relying on brittle scripts, it uses a single-brain exploration loop powered by Gemini to navigate your app, build a permanent coverage map so it never forgets a screen, and automatically document bugs. It outputs a complete HTML report packed with screenshots and exact bounding boxes for every layout flaw, typo, or crash it detects, letting you keep writing code while it handles the QA.

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 artokun/comfyui-mcp-panel

ComfyUI MCP Panel

Build and edit your ComfyUI workflows just by talking to a sidebar AI assistant that actually drives your live canvas. This clever tool embeds a local-first AI agent right inside your sidebar to add, wire, and arrange nodes using natural language, all running directly on your existing Claude or ChatGPT subscription without needing any expensive API keys. You can watch nodes appear as you type, auto-layout messy graphs, and even undo any automated changes instantly with a simple control-Z. It is like having a helpful friend co-pilot your generation pipeline so you can focus on making art.

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 buildermethods/design-os

Design OS

Stop jumping straight into code only to end up with a generic, half-broken app that looks nothing like what you wanted. This tool lets you step back and map out your product vision, sketch your data shape, and design your interface alongside an AI companion. Instead of forcing coding agents to design and write logic at the same time, you build a clear visual and structural blueprint first. When you are done, you export a complete handoff package with props-based React components, TypeScript types, and ready-to-use prompts. It gives your AI coder a perfect target, saving you hours of frustrating debugging.

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 team-reflect/reflect-open

reflect-open

Your notes are just plain files on your own device, and reflect-open keeps them that way while giving you the power of modern note-taking. It is an open-source, offline-first application for Mac and iOS built around a simple folder of Markdown files. Instead of locking your thoughts in a proprietary database, it uses wiki links to connect ideas, shows what notes link back to each other, and opens instantly to today's daily note. You can search notes locally, transcribe audio memos, and even query your notes using your own private AI keys. Best of all, everything stays completely under your control, syncing through iCloud or Git.

📰 https://news.ycombinator.com/item?id=48905524

🆔 @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
📦 blitzdotdev/blitzos

BlitzOS

Your Claude cloud sessions get a fresh virtual machine every single time, meaning every new agent boots knowing absolutely nothing about your setup. BlitzOS solves this by building you a private context repository that instantly teaches new cloud agents your entire development environment. It automatically maps your repositories, aligns your coding conventions, and logs active sessions, letting you launch pre-configured multi-repo workspaces with a single click. Best of all, because it uses Git pointers instead of copying code, your proprietary files never touch outside servers, allowing your cloud agents to push code and open pull requests while your laptop is closed.

🆔 @hackernewsgithubprojects