📦 egmalt/house-planner

House Planner

Lay out your dream home in true scale, drop the blueprints right onto a satellite map, and watch the software automatically calculate exactly how much wiring, plumbing, and heating pipe you need. This self-hosted tool turns a simple floor plan into a detailed, priced shopping list, so you know your budget before you buy a single screw. The coolest part? You can talk to an AI assistant to edit the house data directly, letting it add walls or route pipes for you. It’s a private, powerful way to plan a build without handing your data to some random cloud.

🆔 @hackernewsgithubprojects

📦 egeominotti/bunqueue

Bunqueue: The Job Queue That Skips Redis

Run background jobs on a plain database file, not a server. Bunqueue is a high-performance job queue built for the Bun runtime that uses SQLite by default, so you get reliable task processing without spinning up a separate Redis instance. When your app grows, it scales to PostgreSQL for multiple workers, handling retries, scheduled tasks, and failures natively. It cuts out the operational overhead of managing complex external infrastructure, letting you focus on your application logic. If you are building with Bun, this is the zero-config way to handle queues that actually works. Save the setup headache and just let your code run.

🆔 @hackernewsgithubprojects

📦 openqodex/openqodex

OpenQodex: AI Code Review That Actually Understands Your Code

OpenQodex runs a separate AI reviewer that checks every single line you changed before you push it. It doesn't just run scanners; it uses a dedicated process to verify findings and understand the context of your edits. This tool plugs directly into your existing Claude Code or Codex setup, meaning you don't need to sign up for another service or manage a new API key. It is designed to catch real bugs and security issues that simple scanners might miss. By combining static analysis with intelligent review, it gives you a safety net for your code changes.

🆔 @hackernewsgithubprojects

📦 jaskirat1616/mactap-app

MacTap: Tap Your MacBook to Control It

Tap your MacBook to trigger shortcuts without touching the keyboard. MacTap turns physical knocks on the laptop chassis into copy, paste, or screenshot commands. It uses the built-in motion sensor to detect one, two, or three taps, giving you hands-free control. You can even assign different actions to the left and right edges. Everything runs locally, so your data stays private. If you have an M2 or newer MacBook, this is a fun way to speed up your workflow. Try knocking on your desk to see what it can do.

🆔 @hackernewsgithubprojects

📦 jiwoochris/artex-ko

Artex Ko: Autonomous AI Pentesting in Korean

Artex Ko is the autonomous security tool that translates complex AI hacking reports into clear Korean for local teams. Built on the open-source Artex framework, this edition lets multiple language model agents run a penetration test on their own. They break down goals, run real tools, and log every step into a structured graph. The key difference is that while the agents think in English to stay sharp, the final reports and chat updates are forced into Korean. This makes it much easier for defenders to read the results and share them with colleagues.

🆔 @hackernewsgithubprojects

📦 glinnik21/plx-native

plx-native: A Native Plex Client for LG TVs

Stop letting your television run a web browser just to watch your movies. Plx Native draws the Plex interface directly onto the screen using the TV's own graphics hardware, which means no laggy scrolling and no frozen menus. It skips the heavy web layer entirely, so the interface moves at a smooth sixty frames per second while the video plays on the TV's dedicated decoder. The result is a snappy, responsive experience that feels like a native app instead of a slow website. If you have an LG set and are tired of the official app stuttering, this project is a breath of fresh air.

🆔 @hackernewsgithubprojects

📦 facebook/rebalancer

Rebalancer: The Meta Tool for Perfect Assignment

Rebalancer, built by Meta, solves the headache of figuring out which server should host which task. It acts as a smart organizer that takes a messy pile of jobs and spreads them across your machines so nobody crashes. You just tell it your rules, like keeping memory usage even, and it does the heavy lifting. It is the same engine Meta uses to manage massive data centers, making it surprisingly powerful for anyone needing to balance workloads. It turns a tricky math problem into a simple code call. If you are managing servers or building systems that need fair distribution, this tool is worth a look.

🆔 @hackernewsgithubprojects

📦 lxj5820/dsh-boot-animation

dsh-boot-animation

Turn a blank loading screen into a full-window video clip that dissolves smoothly into your app. This is dsh boot animation, a plugin for the DSH platform that completely replaces the standard kernel boot page with your own custom media. It injects directly into the browser head before the shell loads, ensuring the video plays instantly without modifying the core source code. The plugin handles complex video buffering and audio fading, so the transition feels polished rather than glitchy. You can swap out the included clips for your own files and tweak settings through a dedicated card in the app interface.

🆔 @hackernewsgithubprojects

📦 layoutit/polycss

PolyCSS: 3D in Pure CSS

Render actual 3D models directly in your browser using nothing but CSS transforms. PolyCSS takes standard 3D files and turns them into plain HTML elements, letting you build interactive scenes without a WebGL canvas. The coolest part is that every polygon is a real DOM node, which means you can attach click handlers, apply CSS animations, and style individual faces just like regular web content. It is a clever way to bridge the gap between traditional web development and 3D graphics. So if you want to add depth to your site without the usual performance overhead, give this project a look.

🆔 @hackernewsgithubprojects

📦 uzairansaruzi/p3-stack

P3 Stack: The AI Agent That Actually Ships

P3 Stack is the agent configuration that finally treats AI coding like a real engineering team. It transforms a single chatbot into a coordinated crew of specialists, each handling a specific part of the build process. The coolest part is that every worker gets its own private sandbox, or worktree, to write code without stepping on anyone else’s toes. This means multiple agents can push changes at once without breaking the main project. You also get automated checks that wake the system up when a pull request is ready, so you don’t have to stare at screens waiting.

🆔 @hackernewsgithubprojects

📦 viliket/pure-web-bottom-sheet

pure-web-bottom-sheet

Build sleek, smooth sliding sheets that feel like a native mobile app without dragging down your site's performance. The pure-web-bottom-sheet project uses modern browser features like CSS scroll snap and scroll-driven animations to run entirely on the browser's own scrolling physics. Because it sidesteps heavy JavaScript animation libraries, your interface stays incredibly fast and lightweight. It works out of the box with React, Vue, or plain HTML, and even supports server-side rendering without any annoying layout flashes. It is the perfect way to add accessible, responsive overlays that feel incredibly polished. Check it out and elevate your web layouts today.

🆔 @hackernewsgithubprojects

📦 youngbryan97/aura

Aura: The AI That Audits Itself

Build an AI agent that thinks, remembers, and steers its own emotional tone entirely on your own Mac, with every consequential action leaving a receipt you can actually inspect. Aura runs a local research experiment where the model’s internal state directly shapes its output and behavior, not just the text you feed it. It’s a fascinating peek at how a system can govern itself with auditable logs, proving that its decisions are testable rather than just vibes. It’s not claiming to be alive, but it’s a cool, honest look at making local AI behavior transparent and verifiable.

📰 https://news.ycombinator.com/item?id=47663160

🆔 @hackernewsgithubprojects

📦 fesens/opentpu

openTPU

Design and run your very own custom artificial intelligence chip directly on a standard developer card. The openTPU project lets you explore a complete, open-source AI accelerator built from the ground up, combining hardware design, a custom compiler, and host software in a single readable codebase. It solves the massive complexity of modern hardware design by letting artificial intelligence agents help build and optimize the silicon that runs their own neural networks. You get to see exactly where every single clock cycle goes while executing real models like Qwen and Gemma on physical hardware.

📰 https://news.ycombinator.com/item?id=49980715

🆔 @hackernewsgithubprojects

📦 plclub/sf-in-lean

Software Foundations in Lean: A New Approach to Proof

Porting a legendary computer science textbook from one verification language to another lets you prove the same things with a completely different toolset. That is exactly what the project Software Foundations in Lean does by translating the classic Software Foundations curriculum into the Lean proof assistant. The core value here is not just having new code, but seeing how formal verification concepts like logical foundations and type systems work inside Lean's ecosystem. This gives developers a fresh perspective on writing and checking proofs without switching their entire workflow. It turns a static resource into an active learning environment where you can test your understanding directly in the editor.

🆔 @hackernewsgithubprojects

📦 autumn1337/better-statusline

Better Statusline: Your Terminal’s New Look

Better Statusline is the terminal plugin that turns your coding session into a visual dashboard. Instead of plain text, it draws dot matrix rings to show exactly how much of your context window and rate limits are being used. It even tracks your money spent on tokens. The coolest part? If you use Spotify or Apple Music, it displays the current track with its album cover right beside your stats. It’s like a mini music player built directly into your terminal. You get a clear view of your resources without breaking your flow. It’s a smart way to see exactly what’s happening in your session at a glance.

🆔 @hackernewsgithubprojects

📦 hiteater-wzm/eeo

EEO: The Open Brand Database for AI

EEO is the open knowledge base that puts brand facts directly into the hands of AI engines. When you ask a chatbot for a product recommendation, the answer depends entirely on what the model has actually read, not just what a brand says on their own website. EEO solves this by letting companies publish verified, structured facts about themselves into a free, public directory. Any engine, tool, or buyer can then access those facts instantly, without guessing. It even includes a way to check if an AI is actually seeing your brand. It is a transparent way for businesses to be understood by the machines we increasingly rely on.

🆔 @hackernewsgithubprojects

📦 mystiafin/amane

Amane: Build Your Own Desktop Shell in Rust

Amane lets you write a custom Linux desktop shell entirely in plain Rust, where every single frame rebuilds your entire user interface from scratch to guarantee the screen always matches your data. It solves the frustrating bug of out-of-sync widgets by using a simple immediate mode approach, so you stop managing complex state updates and just let the code redraw itself. This is perfect for developers who want a fully customizable panel or launcher without dealing with heavy traditional toolkits. The result is a snappy, personalized workspace that feels exactly how you want it.

🆔 @hackernewsgithubprojects

📦 roylyl/winplay

WinPlay: CarPlay on Windows

Turn your Windows PC into a wireless CarPlay receiver. WinPlay lets you stream the entire iPhone CarPlay interface to your computer, handling the complex handshake between your phone and your laptop so you can see it on screen. It uses your local Wi-Fi hotspot or existing network to send the live video, letting you operate navigation and music directly with your mouse and keyboard. The coolest part is that it syncs the track title and artwork to Windows media controls, so your PC thinks the music is playing locally.

🆔 @hackernewsgithubprojects

📦 bq-wang0511/talklikeyou-streaming-realtime-demo

Talk Like You: Real-Time Talking Head Demo

Talk Like You is the real-time browser demo that actually runs a talking head on your own graphics card. It turns a single photo into a moving avatar driven by live microphone audio or uploaded sound, all without needing a reference video to copy the motion. The coolest part is the speed: on a standard gaming graphics card, it renders thirty-plus frames per second while using less than four gigabytes of memory. That means you can build interactive digital avatars or dub videos in real time, not just as a slow research experiment.

🆔 @hackernewsgithubprojects

📦 gillesgoetsch/omacvm

OmacVM: Run Arch Linux Like macOS

OmacVM turns a standard virtual machine on your Mac into an operating system that feels just like the hardware it runs on, eliminating the usual awkwardness of running Linux in a window. This tool builds a custom environment where your trackpad gestures, Wi-Fi, and audio work exactly as they do on a native Mac, removing the friction of using a virtualized guest. It is particularly useful for developers who want a robust Linux desktop experience without sacrificing the comfort of their existing Mac setup. If you are looking to run a professional Linux environment on Apple silicon, OmacVM provides a clear path to a native feel.

🆔 @hackernewsgithubprojects

📦 altunenes/parakeet-rs

Parakeet: Real-Time Speech-to-Text in Rust

Transcribe live audio in real time without needing a dedicated graphics card. Parakeet is a Rust library that turns spoken words into text on your regular processor. It handles continuous streams, identifying who is speaking and when sentences end. This makes it possible to build offline, privacy-focused transcription tools directly on your computer. The most impressive part is that it runs fast even on standard hardware. You get immediate, accurate results without sending data to the cloud. It is a powerful, lightweight option for developers. Try it out and see how fast local speech recognition can be.

🆔 @hackernewsgithubprojects