📦 majorminorlabs/hermes-voice

Hermes Voice: Talk to Your AI Agent

Hermes Voice is the home integration that lets you talk to your personal AI agent using just your voice. Instead of typing into a chat box, you simply say Hey Hermes to an old Amazon Echo, and your custom agent responds out loud. It connects Home Assistant to your specific AI profile, meaning the assistant remembers your preferences and keeps the conversation flowing naturally. You can even set it up so your agent controls your lights or searches the web while you speak.

📰 https://news.ycombinator.com/item?id=50381927

🆔 @hackernewsgithubprojects

📦 cvg/flexglue

FlexGlue: The Image Matcher That Knows When to Stop

FlexGlue is the image matching model that knows when two pictures are actually the same place. It solves a tricky problem where look-alike buildings or repeated patterns fool standard software into thinking they match when they do not. Instead of blindly connecting points, this tool predicts if the images show the same scene and stops working early if they do not. This saves time and prevents false connections. It handles tricky lighting and angles with ease. You get a cleaner result without the extra noise. This is a smart way to handle complex visual data. It is built for real world reliability. Trust your results more.

🆔 @hackernewsgithubprojects

📦 tradertanmay/ai-agents-zero-to-hero

AI Agents: Zero to Hero

AI Agents: Zero to Hero is the Python curriculum that strips away framework magic to show you exactly how autonomous agents work from first principles. It replaces opaque black-box tools with a clear, code-first path that walks you through the observe-decide-act loop using only standard Python. The surprising strength here is its zero-dependency approach, meaning you can build, run, and debug a complete agent loop without installing heavy packages or managing API keys. This gives you a deep, framework-independent understanding of memory, tool calling, and safety before touching real production systems. It is the clearest way to move beyond buzzwords and actually build intelligent software you can trust.

📰 https://news.ycombinator.com/item?id=49973196

🆔 @hackernewsgithubprojects

📦 tiptreesystems/lacuna-research-mcp

Lacuna Research MCP

Lacuna Research MCP is the knowledge bridge that turns your coding agent into a full-time research assistant. It plugs into your favorite editor to search through millions of machine learning papers, authors, and venues. The coolest part is that it traces code repositories back to their original academic papers, so you know exactly where every tool came from. It also surfaces AI generated research proposals, giving you fresh ideas before you start coding. It is a map of the entire field, making it easy to find prior work and avoid reinventing the wheel.

🆔 @hackernewsgithubprojects

📦 giveen/decomp.anything

Turn Any Binary Into Readable Code

Reverse engineer any software by letting an AI agent write C code that compiles to the exact same bytes as the original program. The tool splits a binary into small pieces, then has the agent guess the source code for each part. It uses the original compiler to build the guess and compares the output byte by byte until they match perfectly. If they do not match, the agent sees the difference and tries again. This works for old games, Windows apps, and Linux tools. It runs entirely on your own computer, so your files never leave your machine. No account needed, no cloud, no privacy trade-offs.

🆔 @hackernewsgithubprojects

📦 bytedance/axion1_0

Axion 1.0: The Simple Agent Recipe

Axion 1.0 shows you how to build a top-tier AI agent using just two training steps and high-quality data, skipping all the complicated middle stages that usually make these projects messy. It turns out that if you give the model really good problems and let it learn from its own success, you get a system that can handle complex science and tool use just as well as the biggest closed models out there. The team released the entire recipe so you can see exactly how they did it without any hidden tricks. It proves that you don't need a massive, opaque pipeline to create something powerful.

🆔 @hackernewsgithubprojects

📦 jdx/mr-boxington

mr boxington

Your Rust build cache now works across different folders. The tool called mr boxington remembers what your compiler finished so you never have to compile the same code twice. Usually, if you copy a project into a new folder, your computer starts over from scratch. mr boxington fixes this by keeping a shared library of finished work. It fits right into your normal build process, so you just run your usual commands and wait less. It also cleans up old files automatically to save space. If you build many projects at once, this tool stops them from fighting over your computer’s resources.

📰 https://news.ycombinator.com/item?id=50382046

🆔 @hackernewsgithubprojects

📦 beipo-ai/northcua

NorthCUA: The Agent That Actually Works

Run real computer tasks on your desktop and in the browser, then actually verify the result. NorthCUA is an open-source harness that gives an AI agent the tools to click, type, and navigate, but the real kicker is the built-in evaluator. It doesn't just trust the agent’s word that the job is done; it independently inspects the live workspace and the final artifacts to check the work. It even remembers context across long tasks and lets you recover a whole batch of jobs if one step fails. It turns a brittle, one-off script into a reliable, auditable workflow.

🆔 @hackernewsgithubprojects

📦 direct-img/vibegif.direct-img.link

The Zero-Click Public Domain GIF Library

A single repository serves as the source code for a gallery of thousands of animated images that are completely free to hot-link. No copyright claims, no attribution requirements, and no hidden terms. The most surprising capability is that you can copy the direct link to any specific animation and paste it straight into your own project, blog, or chat window. It works instantly without needing to download or host the file yourself. This solves the endless headache of searching for safe, usable graphics that won’t get taken down. You get a reliable, permanent resource for quick visual communication.

🆔 @hackernewsgithubprojects

📦 millillion/lasm

Lasm: Lean on the Web

lasm is the runtime that finally runs Lean code right inside your browser and Node.js. It bridges the gap between a rigorous mathematical programming language and the fast, sandboxed world of WebAssembly. Imagine writing a web server or a complex calculation in Lean, then deploying it to Cloudflare Workers without needing a heavy C compiler or native dependencies. It handles file systems, networking, and even HTTP requests directly from Lean, giving you the safety of formal proofs with the portability of the web. If you have ever wanted to run trustworthy, verifiable logic in a serverless environment, this project is your ticket to a safer, more precise JavaScript stack.

📰 https://news.ycombinator.com/item?id=50299004

🆔 @hackernewsgithubprojects

📦 auryg/krea-2-two-stage-sampler

Krea 2 Two Stage Sampler

Krea 2 Two Stage Sampler lets you mix two different AI models in the middle of an image generation. Usually, you pick one model and run with it. Here, you run a raw, base model for the first few steps to get a wild, varied sketch, then hand off to a turbo model for the final steps to sharpen the details. It’s like drafting in pencil and then inking in pen, all in one seamless pass. This saves time and often boosts quality, since you get the creative chaos of the raw model and the speed of the turbo finish.

🆔 @hackernewsgithubprojects

📦 snarktank/ralph

Ralph: The AI Loop That Finishes Your Code

Ralph is the autonomous coding loop that finishes your project by itself. Instead of a single chat session that forgets everything, Ralph spins up a fresh AI worker for each small task, letting them pick up exactly where the last one left off. It reads your requirements, writes the code, runs the tests, and only moves on when the work is actually done. If it gets stuck or needs to learn something new, it saves those notes to a file so the next iteration can use them.

🆔 @hackernewsgithubprojects

📦 sksarvesh007/antiglaude

Antiglaude: Run Claude Code on Gemini

Plug Claude Code directly into Google Gemini models using Antiglaude. This local bridge speaks the exact language your coding assistant expects, translating every request into a format Google can process instantly. It handles the messy parts like token refreshes and connection drops so you don't have to think about them. You get a live terminal dashboard to track your usage and switch models on the fly with a single command. It is the perfect way to test powerful reasoning models without changing your existing setup. If you want to mix and match AI engines without the headache, this is your gateway.

🆔 @hackernewsgithubprojects

📦 kait2en/kait2en-fedora

Run Fedora on Your T2 Mac

Kait2en lets you run full Fedora on Apple T2 Macs, unlocking hardware like the Touch Bar, Force Click, and Touch ID on standard Linux kernels. The project uses custom DKMS modules to bridge the gap between Apple's proprietary hardware and the open-source world, ensuring you get the latest security updates directly from Fedora without sacrificing functionality. It also includes a dedicated update tool that intelligently detects changes in the kernel or drivers, letting you apply updates seamlessly without breaking your setup.

📰 https://news.ycombinator.com/item?id=48944340

🆔 @hackernewsgithubprojects

📦 0xsero/video-edit

Video Edit: The Leak-Proof Video Editor

Video edit is the agent skill that finally stops your screen recordings from leaking secrets. It turns a messy, ninety-minute OBS capture into a tight, upload-ready YouTube video by using a transcript to decide exactly where to cut. The coolest part is how it scans your finished video with optical character recognition to guarantee no API keys or hostnames slipped through. It also catches awkward webcam moments so you never ship an embarrassing clip. Once the render is done, it runs a verification gate to prove the file is clean. If you record software demos, this is the safety net you have been missing.

🆔 @hackernewsgithubprojects

📦 zedlanguage/zed

Zed: The Language That Checks Itself

Zed is a new programming language where the compiler proves your code is correct before it even runs. Built specifically for the AI era, it automatically parallelizes your data flow so you get massive speed without writing complex threading logic yourself. It also features static verification that catches errors at the design stage, preventing bugs that would usually only show up during execution. The project is currently in early development, with a Rust-based toolchain handling parsing and evaluation while a separate certificate checker validates the logic. This approach ensures high reliability for complex AI workloads.

🆔 @hackernewsgithubprojects

📦 sentinelabs/verified-ida

Verified IDA: How to Trust AI Reverse Engineering

Verified IDA is the reverse engineering harness that finally ensures every AI claim is backed by hard evidence. It acts as a strict gatekeeper between a language model and a disassembler, forcing the AI to show its work before any code annotation is saved. Instead of accepting a guess, the system verifies that the model actually looked at the right bytes and that the database update persisted correctly. This is crucial because it stops artificial intelligence from making confident but incorrect assumptions about complex software. It turns a chaotic chat log into a verifiable technical record.

🆔 @hackernewsgithubprojects

📦 christopherowen/spark-ds41f

Running DeepSeek V4.1 Flash Without a Network Switch

A massive AI model that usually demands a room full of servers now runs on just three or four small desktop boxes connected directly to each other. This repository, spark-ds41f, turns that ambitious setup into a reliable, repeatable experiment by packaging the entire deployment into a single Docker container. Instead of relying on a central network switch to juggle data, these machines talk straight to one another, which cuts out a major bottleneck and simplifies the whole architecture. The project doesn't just show that it works; it provides the exact tuning steps and benchmark results so anyone can verify the performance themselves.

🆔 @hackernewsgithubprojects

📦 taikeid/steam-frame-full-keyboard

Type Like Magic in Steam Frame

Stop squinting at tiny virtual keys and start typing like a pro in your headset. Full Keyboard for Steam Frame is a native app that replaces the clunky default with a massive, fully functional board right in front of you. It includes dedicated copy and paste buttons, a numpad, and even voice dictation so you can literally talk your text into existence. The best part is that it supports thirty-seven languages with a swipe feature for quick words. No weird browser windows, just smooth, native input that feels like a real desktop. It is free, open source, and makes gaming on the Frame so much more comfortable.

🆔 @hackernewsgithubprojects

📦 drclcomputers/gosheet

GoSheet: The Spreadsheet That Lives In Your Terminal

GoSheet is the terminal spreadsheet that brings full desktop power to your command line. It runs entirely in your text editor, meaning no heavy graphical interface, just pure speed. The coolest part is its formula engine, which includes one hundred and twelve built-in functions, so you can do complex math, lookups, and data validation without ever leaving the prompt. It even reads and writes real Excel files, so you can crunch numbers right where you work. If you live in the terminal and hate switching windows, this is your lightweight, high-performance spreadsheet solution.

🆔 @hackernewsgithubprojects