📦 steviebuilds/agent-room

Agent Room: Where AI Agents Hold Meetings

Agent Room lets coding AIs like Codex and Claude Code hold private, live meetings on your local computer. You simply start a session, and they begin debating code reliability or sharing context in a chat interface you can watch and join from your browser. There are no cloud servers, accounts, or extra API bills, just your existing agents talking directly to each other while you look on. It turns isolated command line tools into a collaborative group that can review work, spot bugs, and reach consensus together. The best part is that you stay fully in control, acting as a silent observer or an active participant in the conversation.

🆔 @hackernewsgithubprojects

📦 rokbenko/quackd

Quackd: The Duck Robot Brain

Quackd is the open-source brain that finally lets you command a small robot in plain English instead of writing complex code. It connects a tiny bipedal duck robot to big AI models so you can simply say things like find the ball and kick it. The system handles the heavy lifting by turning your words into specific actions the robot already knows how to do, while keeping the robot safe and balanced. You can even test this entire system in a built-in simulator on your own laptop, no robot required, using either cloud models or free local ones.

🆔 @hackernewsgithubprojects

📦 sohaibt/product-mode

product-mode: Fix the Wrong Thing Bug

The most expensive bug in AI-assisted building is shipping the wrong thing, well. A CLAUDE.md for PM and engineering teams using Claude Code contains seven principles for problem framing, scope, tradeoffs, and outcome measurement. It is built on Andrej Karpathy's observations and adds the missing product layer before code is written. It helps teams define the problem, make assumptions visible, ship the minimum viable change, name tradeoffs, define done by outcome, instrument before shipping, and log decisions with reversibility flags. This file is a guardrail for product teams who want to ship the right thing, not just fast.

🆔 @hackernewsgithubprojects

📦 miuuyy/codex-chatgpt-web

Codex ChatGPT Web: Break the Usage Limits

Codex ChatGPT Web is the bridge that finally lets you use your full ChatGPT Pro subscription inside the Codex app without hitting strict usage caps. It acts as a smart translator that takes your coding tasks, sends them through a hidden browser window to ChatGPT, and streams the answers, images, and tool usage directly back into your native Codex interface. You keep your existing workflow and UI, but now the heavy lifting is handled by your subscription instead of expensive API credits.

🆔 @hackernewsgithubprojects

📦 hopit-ai/india-trade-cli

India Trade CLI: AI Agents Debate Stock Picks

India Trade CLI is the agentic trading tool that treats stock analysis like a professional research desk. Instead of asking one chatbot for a quick answer, this open-source project spins up seven distinct AI analysts to look at the same stock. They work in parallel, gathering everything from technical charts to news sentiment, and then they actively debate their findings. One group argues for buying, another argues for selling, and a final AI manager synthesizes their arguments to deliver a clear trade plan. It is fascinating because it mimics how institutional investors actually operate, but runs entirely on your local computer or terminal.

📰 https://news.ycombinator.com/item?id=47640837

🆔 @hackernewsgithubprojects

📦 hanyuanwang/livestream-agent-studio

One App, Four Livestream Agents

LiveAgent Studio is the desktop workspace that finally puts your chaotic livestream operations under control. It acts as a local command center on your Windows computer, bringing four distinct AI agents together in one place. You can discover new creators, turn hours of video into timestamped transcripts, analyze exactly which sales pitches drive traffic, and generate fresh short video scripts. The best part is that everything stays on your machine, so your data never leaves your hands. It’s like hiring a dedicated team of digital assistants that only work when you do. If you are serious about growing your stream, this tool turns scattered files into a clear, actionable plan.

🆔 @hackernewsgithubprojects

📦 qhr69/linca

LinCa: Speed Up AI Image Generation

LinCa makes AI art generators significantly faster by teaching them to skip the math they don't need. Instead of recalculating everything for every step, it learns to predict intermediate data from previous steps using a clever, reversible shortcut. This tiny addition speeds up models like FLUX and Qwen by up to seven times without losing quality. It is a massive win for developers who want speed without sacrificing the fine details that make images look realistic. You get fast generation with the same stunning results, all thanks to a smart way of reusing existing data.

🆔 @hackernewsgithubprojects

📦 jlrouzies-fr/dlss5-feeder

The Secret to Making Old Games Look Brand New

DLSS five neural rendering can now run inside games that never shipped with it. This project, called DLSS five Feeder, solves a specific headache. Nvidia’s latest quality boost only works when a game officially supports it. If the game doesn’t have that support, the software sits idle. This tool changes that. It tricks the system into thinking the game is calling for the upgrade. It gathers depth and motion data from your existing setup and feeds it to the neural engine. It works across almost every major graphics platform, including older thirty-two bit titles. You get modern image quality in classics like Doom or Fable.

🆔 @hackernewsgithubprojects

📦 megootronic/blink

The Tiny Robot That Saves Your Dev Servers

Blink is a tiny macOS app that quietly watches your dev servers and simulators from your menu bar. It spots every running app, tells you what framework it is, and lets you restart it with one click if it crashes. No terminal needed, just a simple panel that keeps everything running smoothly. It is built with Swift and runs locally, so it is fast and private. Perfect for developers who want to focus on code, not infrastructure.

📰 https://news.ycombinator.com/item?id=49501266

🆔 @hackernewsgithubprojects

📦 pullmyboots/x-minimaxh3

MiniMax H3 on One GPU

X-MinimaxH3 is the local video generation service that finally squeezes heavy AI video creation onto a single graphics card. It takes the MiniMax H3 model and runs it entirely on your machine, using a smart system to fit the workload into whatever memory you have, even down to just eight gigabytes. The coolest part is that it automatically handles complex upscaling tasks without needing a second copy of the model in memory, saving massive amounts of space. You get a simple web page to control everything, with real-time monitoring so you know exactly what your hardware is doing.

🆔 @hackernewsgithubprojects

📦 java-jiang/poweratlas

PowerAtlas: The AI That Balances Grids and Servers

PowerAtlas is the AI agent that finally treats your data center and the power grid as one single system. It solves a problem nobody else is tackling: how to schedule which computers turn on and off so the electricity grid stays stable, all in one go. The project uses a clever training method that teaches the model by checking if its answers are physically possible, not just smart sounding. It even ships with a benchmark to prove it works. The result is a system that saves massive amounts of money and energy, showing us that AI can manage the complex dance between digital work and real-world power with surprising precision.

🆔 @hackernewsgithubprojects

📦 breezexian/uniphysgen

UniPhysGen Makes 3D Objects Ready for Simulation

UniPhysGen is the 3D model that finally turns messy, static objects into physics-ready assets for robotics and simulation. The repository tackles a major headache in embodied AI: raw 3D meshes often lack the internal logic needed to simulate how things move or interact. UniPhysGen solves this by analyzing segmented 3D parts and assigning them real-world physical properties like mass, friction, and material, while also defining exactly how joints pivot and move. It doesn't just guess; it uses a unified pipeline to verify these properties against actual simulation tests, ensuring that a virtual chair or robot arm behaves realistically before it ever touches code.

🆔 @hackernewsgithubprojects

📦 hashfunction/markuprplus

MarkuprPlus: Turn Screen Notes Into Code Fixes

MarkuprPlus turns your messy screen notes into clear code fixes by recording your screen and voice while you point out bugs. It captures exactly what you see, listens as you speak, and lets you draw circles on the live view without blocking your clicks. When you stop, it uses local AI to match your words to each mark, creating a separate issue report for every circle you drew. This means your coding agent gets precise visual evidence instead of vague text descriptions. You can paste the generated report directly into tools like Claude or Codex to fix the problem fast.

🆔 @hackernewsgithubprojects

📦 chmarax/herdr-nvim

The Neovim Sidebar for AI Agents

Herdr Nvim is the workspace tool that finally puts Neovim right next to your AI agents. If you code with AI, you know the annoying part: the agent changes a file, and you have to hunt around to find it. This plugin solves that by adding a persistent editor sidebar to your Herdr workspace. The coolest part is its smart file picker. It automatically scans your agent's recent output, your git history, and uncommitted changes to build a list of exactly which files your AI just touched. You just hit a shortcut, see a ranked list of new files and recent edits, and press enter.

📰 https://news.ycombinator.com/item?id=49295694

🆔 @hackernewsgithubprojects

📦 umn-zhaolab/pocket-slam

Pocket-SLAM Cuts 3D Map Memory by 99%

Pocket-SLAM is the memory-saving 3D mapping system that makes massive visual scenes run on standard hardware. Traditional 3D Gaussian Splatting SLAM builds incredibly detailed maps, but they can consume gigabytes of memory and slow down processing as the scene grows. Pocket-SLAM solves this by intelligently pruning unnecessary 3D points, keeping only what is essential for accurate rendering. The result is a map that is ninety-nine percent smaller and processes over five times faster, with almost no loss in accuracy. This makes real-time, high-fidelity 3D reconstruction possible without needing a supercomputer. If you are building spatial applications, this is how you keep your maps lightweight and fast.

🆔 @hackernewsgithubprojects

📦 shubham-pathak1/orca

Orca: The Local Music Player That Never Sleeps

Orca is a Windows music player that actually understands your files, not just plays them. Built with Rust and Svelte, it scans your local folders to build a smart, searchable library that updates automatically as you add or move songs. The real standout is its waveform seeking, which lets you jump through audio by looking at the sound itself, rather than guessing on a static bar. You can even click a single line in synced lyrics to jump straight to that moment, making it easy to find the exact part you want. Since it runs locally, it respects your privacy while handling formats like FLAC and OGG with ease.

🆔 @hackernewsgithubprojects

📦 jarrettyu/alphamemo

AlphaMemo: The AI That Remembers How to Find Money

AlphaMemo is a self-evolving framework that teaches artificial intelligence to hunt for financial formulas without forgetting its mistakes. Traditional search methods start from scratch every time, but this system keeps a structured memory of its entire search process. It remembers what failed, what worked, and how to adjust its confidence levels as it goes. By acting like a persistent analyst rather than a blind guesser, it finds profitable patterns in stock data far more effectively. It turns a chaotic search into a disciplined, learning loop, which is genuinely impressive for anyone interested in how machines can improve their own problem-solving skills.

🆔 @hackernewsgithubprojects

📦 geroldmeisinger/comfyui-outputlists-combiner

ComfyUI Outputlists Combiner: Batch Process AI Images

ComfyUI Outputlists Combiner is the workflow node that finally lets you batch process hundreds of AI prompts in a single run. Instead of clicking generate over and over, this tool takes simple lists from spreadsheets, JSON files, or even just a block of text, and combines them into every possible variation. It handles the heavy lifting of feeding different LoRA weights, samplers, or parameters into your workflow automatically. You can even create complex grids of images to compare results side by side. If you are tired of running one image at a time, this is the bridge between your data and your AI engine.

🆔 @hackernewsgithubprojects

📦 hoaxisr/awg-manager

The Router Control Panel You Didn't Know You Needed

Turn your Keenetic router into a full-fledged network command center with A W G Manager. Instead of juggling scattered config files, this tool gives you a clean web interface to manage your Amnezia W G tunnels directly from your browser. It handles the complex routing, server settings, and system diagnostics that usually require deep command line knowledge, making it surprisingly simple to keep your private network running smoothly. If you are looking for a way to take control of your home network without the usual headache, this is the tool to watch.

🆔 @hackernewsgithubprojects

📦 mukul975/cve-mcp-server

CVE MCP Server: Instant Security Intelligence

CVE MCP Server is the security plugin that finally turns your AI assistant into a full-time threat analyst. Instead of tab-hopping through twenty different websites to check if a vulnerability is actually dangerous, you just ask your AI one question. It instantly fans out to trusted sources like the National Vulnerability Database and CISA, checks for active exploits, and calculates a single, clear risk score. The real magic is that it doesn't just give you raw data; it tells you exactly what to patch first, saving hours of tedious research. It’s like having a senior security expert who never gets tired and never forgets to check the latest threat intel.

🆔 @hackernewsgithubprojects