143 subscribers
10.2K videos
10.8K links
Download Telegram
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ opsafari/hypoarena

Hypoarena: How to Prove Your AI Research Isn't Just Vibes

Test whether an artificial intelligence actually reasons by running a full scientific experiment entirely offline. Hypoarena builds a fake library of papers with hidden causal links and forces the system to debate, rank, and evolve its guesses. The surprising part is that it uses a tournament score to verify the machine is finding the real answers. You get a clear report proving exactly where the logic holds up and where it fails. It is not a data leak detector or a simple chat bot. It is a rigorous, repeatable way to audit the thinking process of any model you build, giving you total confidence in your results.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ moonbai-studio/tungsten-edge

Tungsten Edge: The Mac Dock Finally Gets Windows

Tungsten Edge is the macOS taskbar that finally makes window switching feel like a snap. If you have ever clicked a Dock icon just to hunt for the one specific window you needed, you know the pain. This app replaces the default Dock with a bar where every open window gets its own clickable card. No more hunting through menus or guessing which window is which. It is built in native Swift, so it is light, fast, and blends right into your desktop with a frosted glass look. It handles multi-window apps perfectly, keeping your workflow clean without the clutter.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ circuitlord/circuitlordvrmodinstaller

CircuitLord's VR Mod Installer: One Click to VR

CircuitLord's VR Mod Installer is the simple desktop app that turns complex virtual reality gaming setup into a single click. Instead of hunting through forums for file zips and mod loaders, this tool automatically finds your Steam games, downloads the necessary plugins, and installs full virtual reality support for titles like Big Walk and Titanfall 2. The most impressive part is its precision; it tracks exactly which files it places, so updating or uninstalling the mod never breaks your original game files.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ iamccs/iamccs-nodes

IAMCCS Nodes

IAMCCS Nodes is the ComfyUI extension pack that finally makes AI video production feel like a real movie studio instead of a tangled mess of wires. You know that annoying bug where you plug in a style filter and it just doesn't show up? This pack fixes that by forcing the model to actually use your settings. It acts like a director's board, keeping your shots, audio, and timing perfectly in sync so your clips flow together naturally. If you are tired of fighting your software, this is the clean, organized way to build high-quality video sequences that actually look good.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ addyosmani/web-quality-skills

Web Quality Skills: The Missing Instruction Set for AI Coding Agents

Your coding agent can write code, but it often guesses how to make that code fast. Web Quality Skills fixes this by giving agents a set of instruction files that teach them to use real browser data, not just code reading. It instructs the agent to run live performance tests and check actual user experience metrics before suggesting changes. This ensures every fix is backed by evidence rather than assumption. It is the difference between an agent that guesses and one that knows. If you use AI to build websites, this project makes sure it actually understands how to make them work well for real people.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ opendcai/datamind

DataMind: The AI Assistant That Remembers

DataMind is an AI assistant that actually remembers what you tell it while you talk. Most chatbots forget everything the moment you ask a new question, but DataMind has a built-in memory that writes down new facts and uses them instantly in your very next reply. It is not just a chat bot; it is a shared data plane that lets the agent store and retrieve information like a document, a database row, or a relationship between people. This means it can track your preferences, recall specific details from your files, and answer complex questions with real evidence.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ soundpulse/hermes-live2d

Hermes Live2D: Your Agent's New Face

Speak through your computer by turning a simple coding agent into a living desktop companion. This project gives a remote AI a face and a voice, letting it play recorded audio with real lip sync while floating on your screen. The coolest part is how it connects a distant server to your local Mac, allowing the character to track your mouse and display speech bubbles right from your agent's text output. It’s a neat way to see your automation actually talk, making the digital interaction feel personal instead of just reading logs. Give your next script a personality and watch it come alive.

πŸ“° https://news.ycombinator.com/item?id=48887243

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ 0xethanq/astra-quant-agent

AstraQuant: The AI Trading Team With a Safety Net

Place crypto trades autonomously using a team of A.I. models that debate strategy before a C.I.O. agent makes the final call. This project, called AstraQuant, solves the problem of trusting black-box algorithms by adding a physical safety layer. It runs on the O.K.X. exchange and uses strict Python rules to block any risky move, ensuring no human or bot can accidentally blow up your account. The coolest part is that if the system isn't one hundred percent sure, it simply refuses to trade. It is a transparent, self-correcting trading desk that you can actually read and control. If you want A.I.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ lotchuazzz-crypto/papergraph-mcp

PaperGraph MCP: Stop Guessing Math Papers

PaperGraph MCP turns dense math papers into clear reading maps for AI agents. Instead of guessing which results matter or where proofs hide, it extracts key theorems and traces the exact evidence supporting them. You can see precisely where every claim comes from in the original text, then plan a logical reading order without getting lost in citations. It also helps you review external dependencies so you know what else to read before diving deep. This is huge because it replaces vague summaries with grounded, verifiable insights. If you work with academic math, this tool gives you a structured way to actually understand the work. No more guessing.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ 555cute/astra-quant-agent

AstraQuant: AI Trading with Human Oversight

AstraQuant is the autonomous trading terminal that lets AI models debate while hard-coded Python code holds the veto power. It connects directly to the OKX exchange, where a committee of artificial intelligence agents analyzes market data and proposes trades. But here is the part that makes it safe: before any order actually executes, it must pass through a physical risk pipeline that can reject bad ideas instantly. This ensures that while the AI handles the strategy and thinking, the actual money management is locked down by strict, readable code you can inspect.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ fsiaonma/elpis

Elpis: The AI-Ready Full-Stack Framework

Elpis is an enterprise-grade full-stack framework that quietly integrates advanced AI agents, memory, and retrieval tools directly into your standard NestJS backend. Instead of building these complex features from scratch, you define your data structure once, and the framework automatically generates the entire admin dashboard interface for you. This means you get a professional-looking management UI, complete with search bars, data tables, and edit forms, just by describing your database fields in a configuration file. It’s a clever way to skip the repetitive work of writing frontend code for internal tools.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ tautvydasderzinskas/thingport

Thingport: Your Personal 3D Library

Thingport lets you pull 3D models directly from major printing sites into one private library that runs on your own hardware. It solves the problem of your favorite files getting lost in random folders or scattered across different websites. You can grab a model with a browser extension, see a clear 3D preview right in your browser, and keep your notes and print settings in one place. It is a self-contained tool that gives you full control over your digital collection. If you print, this is the simple way to keep your files organized and accessible.

πŸ“° https://news.ycombinator.com/item?id=49684004

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ callbacked/kinesis

Control Your Mac With Your Brain? Well, Your Wrist

Kinesis is the menu-bar app that turns your Meta Neural Band into a native mouse for your Mac. It lets you swipe between desktops, open Mission Control, and adjust volume by pinching and twisting your wrist in the air. The coolest part is how it handles movement. It uses your forearm position to move the cursor and your pinches to click. But your arm naturally sways while you hold it still. Kinesis figures out exactly how much your arm trembles and ignores that specific amount of shake. So the pointer stays perfectly locked in place, even if you are just hanging out.

πŸ“° https://news.ycombinator.com/item?id=49695408

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ openclaw/openclaw-enterprise

OpenClaw Enterprise: Kubernetes for AI Agents

OpenClaw Enterprise lets you deploy and manage AI agents exactly like you manage containers, turning messy, one-off scripts into a governed, scalable system. It acts as a control plane, handling everything from spinning up secure sandboxes to managing credentials and monitoring performance, so agents don't just run; they operate reliably. Whether you are using Docker or Kubernetes, this platform abstracts the complex infrastructure, letting you focus on what your agents do, not how they are hosted. It is built for teams that need strict access controls, detailed auditing, and the ability to scale workloads without rebuilding the foundation.

πŸ“° https://news.ycombinator.com/item?id=49901210

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ andretripol/openghost

OpenGhost: The Desktop AI That Actually Builds

OpenGhost is the desktop AI agent that actually shows you the work while it happens. Instead of just spitting out text, it draws live charts, flowcharts, and diagrams right next to its explanations so you can visually verify the logic before you trust it. It even includes a built-in browser that lets the AI click, type, and navigate the web on its own, all while keeping your data strictly on your local machine. It is a fully open-source desktop app that feels like a collaborative partner rather than a black box.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ thu-nics/tah

Think-at-Hard: The LLM That Skips Easy Work

Think-at-Hard makes language models smarter by spending extra effort only on the hardest parts of a problem. Instead of re-evaluating every single word it generates, this new method uses a tiny internal checker to spot difficult tokens and refine just those. It skips the boring, easy ninety-four percent of the work entirely. By focusing its brainpower where it matters most, the model actually gets better at math and logic than standard models, even while doing less computation. It is a clever trick that proves you do not need to slow down to think deeply.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ xhtoken/spark-x2.5

Spark-X2.5: Local AI Power

Run powerful AI agents directly on your own hardware without relying on cloud servers. Spark-X2.5 delivers serious coding and reasoning skills in a compact model that fits on personal devices. The standout feature is a unique attention design that lets the model process one million tokens of context efficiently. This means you can feed it entire codebases or long documents without slowing down or losing track. It solves the problem of needing massive, expensive data centers for everyday intelligent tasks. By making high-level AI accessible and local, it gives developers full control over their data and workflows. The result is a practical tool for anyone building smart applications locally.

πŸ“° https://news.ycombinator.com/item?id=49907193

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ fastcrw/crw

crw: The Tiny Rust Web Scraper

A single Rust binary that runs in just six megabytes of memory is outpacing massive cloud services like Tavily and Firecrawl by more than twice in speed. This project is a lightweight web engine that turns any website into clean markdown or structured data for artificial intelligence agents. It solves the problem of slow, heavy scraping tools by providing a fast, local alternative that speaks a simple language your coding assistants already understand. You can run it on your own machine without a server, or use their cloud version.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ mafifrizi/ares

Ares: The Automated Red Team Platform

Ares calculates the shortest path to total network compromise before a single packet is sent. This platform automates authorized red team engagements by mapping multi-stage attack paths across active directory environments, proving exploitability rather than just listing vulnerabilities. It enforces strict scope boundaries to prevent accidental out-of-scope access while storing harvested credentials in an encrypted vault. Security teams use it to validate defensive posture continuously without the cost of manual testing. The result is a clear, deterministic view of how an attacker would move through the network. This shifts security validation from guesswork to precise, auditable proof of risk.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ posthog/jeeves

Jeeves: The Model That Thinks Before It Decides

Jeeves is a nine billion parameter decision model that actually pauses to think through a problem before giving you an answer. Built by PostHog, it solves a common issue where standard classifiers guess quickly but often get it wrong. By training a specific base model to reason through its logic, Jeeves achieves higher accuracy than its competitors even on new data it has never seen. It handles yes or no questions, multiple choice, and ratings all in one request while remaining fast enough for real time use. If you are tired of low accuracy in automated decision pipelines, this is a clear upgrade that proves thinking before answering pays off.

πŸ“° https://news.ycombinator.com/item?id=49891290

πŸ†” @hackernewsgithubprojects