📦 aim-intelligence/video2robot

Video2Robot

Convert human movement from any generative video directly into physical humanoid robot motions. You just type a prompt to generate an AI video of a movement using Google Veo or OpenAI Sora, or upload a standard clip of a human doing a backflip, walking, or dancing. The software automatically handles the entire pipe, extracting the precise three-dimensional human pose frame-by-frame, and immediately retargets those joint angles to match physical humanoids like the Unitree G1. It completely bypasses manual hand-animation by translating visual actions into raw motor commands. Give it a spin to watch your robot mimic any video instantly.

🆔 @hackernewsgithubprojects

📦 mattpocock/dictionary-of-ai-coding

Dictionary of AI Coding

The average AI coding model doesn't actually forget your instructions over time; instead, the signal simply gets drowned out by background noise as your chat history grows. The dictionary-of-ai-coding repository is a brilliant, plain-English glossary that translates confusing AI jargon into practical developer reality. It defines terms like attention degradation and vibe coding by focusing on the real-world symptoms you actually encounter at your keyboard. Instead of stacking academic fluff, it explains exactly why your AI assistant suddenly starts ignoring your code schema or making up fake variables mid-session. It is the perfect handbook for any developer wanting to understand how these tools think so they can write better code...

📰 https://news.ycombinator.com/item?id=48794864

🆔 @hackernewsgithubprojects

📦 alookai/alook

Alook

alook is the self-hosted collaboration platform that coordinates your AI agents using standard email. Instead of hiding your automated workflows inside closed databases, this system gives your digital assistants their own inboxes to communicate with each other and with you. Agents can share memory, coordinate tasks, manage calendars, and run skills directly from your local hardware using a dedicated desktop app and command-line daemon. It turns your background scripts into actual teammates you can cc on a thread, creating a transparent, auditable history of how your automated work gets done. It is a highly practical way to build a cooperative AI workforce on your own terms.

🆔 @hackernewsgithubprojects

📦 ammaarreshi/generals-mac-ios-ipad

generals-mac-ios-ipad

generals-mac-ios-ipad is the open-source porting project that brings Command and Conquer Generals Zero Hour natively to modern Apple devices. Instead of slow emulation, this project recompiles the original 2003 game engine to run natively on Apple Silicon. It translates the original graphics pipeline to Metal and adds intuitive touchscreen gestures designed specifically for real-time strategy games on your iPad or iPhone. By pairing your own legally owned game files with this project, you can play full skirmishes and campaigns on the go. It shows how classic PC gaming can live on natively through modern engineering.

🆔 @hackernewsgithubprojects

📦 productdevbook/hucre

Hucre Spreadsheet Engine

Hucre is the lightweight spreadsheet engine that finally lets you read and write Excel, CSV, and OpenDocument files with zero dependencies. It packs streaming, styling, password encryption, and even legacy binary file support into a tiny eighteen-kilobyte package that runs perfectly in any browser or cloud environment. While massive libraries slow down your application, this tool uses pure TypeScript to handle large datasets row-by-row without eating up your memory. You can even edit spreadsheets without losing existing charts, macros, or pivot tables. It is the perfect tool for fast, clean, and modern data processing.

🆔 @hackernewsgithubprojects

📦 kanavtwtgg/birds.cafe

birds.cafe

Soar over a vast, procedurally animated ocean to escape the stress of daily life. This calming browser simulator gives you direct control over a flock of seagulls in V-formation, letting you steer them through gorgeous day and night cycles, passing rainstorms, and sudden lightning. There are no high scores, time limits, or objectives here, just relaxing ambient music and beautiful physics-based flight physics. Built using WebGL and atmospheric scattering shaders, it runs beautifully on both desktop and mobile screens. If you need a quick mental break, it is the perfect online sanctuary to sit back, glide with the wind, and just watch the water pass by below.

🆔 @hackernewsgithubprojects

📦 ircss/blender-geometry-node-french-houses

Blender Geometry Node French Houses

Build an entire historic European town in Blender just by drawing basic shapes. The blender-geometry-node-french-houses project gives you a complete toolset of highly organized geometry nodes to procedurally generate northern French style architecture. Instead of manually modeling every stone, you can define the simple foundation of a house in edit mode, and the system automatically constructs the walls, roofs, and windows. It even handles complex structures like gothic cathedrals, round towers, and winding wooden bridges. Grab this project file to effortlessly spin up detailed historic villages and study how a master artist structures clean, readable node trees.

🆔 @hackernewsgithubprojects

📦 jia-ethan/pavedpath-code

PavedPath Code

Find proven solutions to your hardest coding errors by tapping directly into open-source history. PavedPath Code is a reusable skill designed for AI agents that extracts real, verified implementation paths from GitHub repositories, merged pull requests, and active issue threads. Instead of generating guesses or writing code from scratch, it searches public developer evidence to see how other engineers successfully solved the exact same framework issues, dependency blocks, or runtime errors. It then translates those proven open-source patterns into a minimal, safe adaptation for your local codebase, leaving you with a highly reliable fix and a clear way to verify it.

🆔 @hackernewsgithubprojects

📦 mudler/locate-anything.cpp

locate-anything.cpp

The C++ inference port that finally makes high-speed object detection accessible on everyday hardware, locate-anything.cpp is a lightweight alternative to bloated Python runtimes. It takes NVIDIA's LocateAnything visual model and ports the entire pixel-to-labeled-boxes pipeline into C++ using a simple command-line interface. Instead of wrestling with massive dependencies and heavy GPU requirements, you get exact, bit-for-bit identical detections on your CPU up to five times faster. It is perfect for developers who want to pass an image and a plain text prompt to get instant, labeled bounding boxes without the setup headache. Check it out and start detecting things locally today.

🆔 @hackernewsgithubprojects

📦 forgetmeai/freedeepseekapi

freedeepseekapi

You can now run DeepSeek Web Chat as a completely free, local API server without paying for official API tokens. A clever new tool called freedeepseekapi routes your standard development requests through a real, logged-in browser session on your computer. It acts as a local proxy that translates standard OpenAI and Anthropic format calls directly into DeepSeek web chat interactions. This means you can plug the free web interface straight into popular tools like Open WebUI, Claude Code, or your own local scripts, complete with streaming, web search, and reasoning outputs. It even handles multi-account pools to prevent rate limits. Just log in once, boot up the server, and enjoy...

🆔 @hackernewsgithubprojects

📦 cursortouch/windows-use

Windows Use

Windows Use is the automation library that lets AI agents navigate your desktop with precision. Instead of relying on clumsy computer vision or taking random guesses with coordinates, it connects directly to the underlying Windows accessibility framework to read and manipulate interface elements natively. It gives your models a structured, real-time map of every button, text box, and menu on the screen. This means your agents can click, scroll, and type with absolute accuracy, even if windows are overlapping or off-screen. It is a brilliant way to build digital assistants that act like real power users.

🆔 @hackernewsgithubprojects

📦 schlae/cm5-reveng

cm5-reveng

Snoop around the inner workings of a Raspberry Pi Compute Module 5 without destroying your own hardware. The cm5-reveng project is a fascinating, fully reverse-engineered schematic and board layout of the popular system-on-module. By meticulously sanding down a module layer by layer and scanning it at ultra-high resolution, a hardware designer reconstructed its entire ten-layer board and mapped out all those mysterious, undocumented test points. While it is not meant to be manufactured due to custom chips and signal complexities, it is pure gold for advanced debugging, educational hacking, or salvage. Grab the schematics and start exploring how the hardware really functions under the hood.

🆔 @hackernewsgithubprojects

📦 yynxxxxx/codex-x

Codex-X

Codex-X is the desktop manager that finally makes customizing your Codex workflows incredibly easy. Instead of struggling with terminal commands and manually editing configuration files, this tool gives you a clean visual dashboard to manage everything in one place. You can switch between different API providers with a single click, instantly edit your configurations, and manage your official login credentials. Its most exciting feature is the built-in prompt manager, which lets you instantly write unrestricted testing templates into your setup for advanced security research like SQL injection and reverse engineering. It is the perfect control center for anyone looking to get more out of Codex without the configuration headaches.

🆔 @hackernewsgithubprojects

📦 oomol-lab/open-connector

open-connector

Your AI agents can now talk directly to over one thousand popular apps and services with a single open-source tool. The open-connector repository is a lightweight authentication gateway that bridges the gap between smart AI helpers and the massive world of online software. Instead of writing custom API code and wrestling with complex login screens for every single app, you use this project to handle the security and token refreshing in the background. It turns complex services into simple tools your digital assistant can use right away. It is the ultimate translator for AI communication.

🆔 @hackernewsgithubprojects

📦 augani/openreel-video

OpenReel Video

The browser-based video editor that finally brings professional, zero-install editing right to your web browser without holding your data hostage. OpenReel Video is a fully-featured, open-source editor that runs entirely on your local machine. It uses WebGPU and hardware-accelerated WebCodecs to handle heavy tasks like multi-track timeline editing, keyframe animations, and 4K exports right inside Chrome or Edge. Because everything processes client-side, you never have to upload giant files to a random cloud server, and you do not have to worry about subscription fees or annoying watermarks. It is a brilliant way to edit videos quickly, privately, and for free.

🆔 @hackernewsgithubprojects

📦 ruzin/stenoai

Steno

Record, transcribe, and summarize your confidential meetings completely on your own device using local artificial intelligence. Steno is a privacy-first notepad app that keeps your sensitive discussions entirely secure by running everything offline on Windows or Mac. It captures system audio from virtual meetings, generates live transcripts with speaker labels, and crafts structured markdown summaries without ever sending a single byte of your data to the cloud. You get full control over your meeting notes while using highly efficient local models. Download it today to keep your conversations truly private.

🆔 @hackernewsgithubprojects

📦 kotohibi/3dgs_pipeline_guide

3dgs_pipeline_guide

You can transform raw 360-degree video footage into incredibly sharp, highly accurate 3D Gaussian Splats with minimal manual effort. The 3dgs_pipeline_guide repository lays out a brilliant walkthrough for turning panoramic captures into stunning virtual environments. It tackles the biggest headache of 3D modeling by utilizing specialized software to directly align spherical images, saving you from tedious manual alignment. The absolute best part is how it uses an automated utility to extract only the sharpest video frames while instantly masking out moving cars and people so they do not ruin your final scene. It is a fantastic guide for anyone wanting to easily capture the real world in immersive three dimensions.

🆔 @hackernewsgithubprojects

📦 liunian06/ko-lesson

ko-lesson

College students can now convert messy, chaotic class slides and PDFs into a structured study system for Obsidian. Instead of just summarizing documents, ko-lesson works as an interactive study assistant that maps out a customized learning path based on your real course materials. It reorganizes dense slides into readable markdown files, builds concept flashcards, tracks your weak points, and even generates practice exams with customized follow-up tests for your mistakes. Whether you want to study one lesson at a time with active feedback or generate a massive exam cram package overnight, it transforms confusing materials into an organized workspace you can actually use to ace your finals.

🆔 @hackernewsgithubprojects

📦 nerd-lang/nerd-lang-core

Nerd Lang Core

Write backend logic with fifty percent fewer tokens using a language designed for AI models instead of humans. That is exactly what nerd-lang-core does. Since machines write nearly half of our code today, this project ditches braces, semicolons, and complex syntax in favor of plain English words that tokenize perfectly for LLMs. This drastically cuts down token usage and lowers API billing costs while keeping everything easily auditable by humans. It even compiles straight to native code using LLVM, giving you lightning-fast performance alongside built-in agent capabilities like direct Claude API calls and HTTP requests. Let the bots write the code and save a fortune on your next LLM project.

🆔 @hackernewsgithubprojects

📦 christopherkarani/espresso

Espresso

You can now run and train machine learning transformers directly on Apple's Neural Engine, bypassing CoreML entirely to get a massive speed boost. A clever project called espresso uses reverse-engineered private Apple APIs to compile models straight to the silicon. By staying completely on the hardware and avoiding the overhead of standard Apple frameworks, it runs token generation up to four times faster than CoreML. It utilizes zero-copy memory buffers and compiles the model once to reuse across steps, which stops the processor from constantly re-analyzing the network. It is a brilliant way to unlock the true potential of your Mac.

📰 https://news.ycombinator.com/item?id=48794803

🆔 @hackernewsgithubprojects