📦 qwen-applications/skill-self-play

Skill Self Play

Skill Self-Play is the training framework that finally helps artificial intelligence models teach themselves advanced logical reasoning and tool-calling skills. Instead of feeding models a fixed, unchanging set of training tasks, this repository lets an AI questioner and an AI solver play against each other to co-evolve. The questioner generates fresh, verified tasks tailored specifically to the solver's current learning limit, pushing the boundaries of what the model can actually handle. It makes training highly targeted, adapting as the model grows smarter. You get vastly stronger reasoning capabilities at training time, while your final model stays simple and fast for everyday use.

🆔 @hackernewsgithubprojects

📦 lyl1015/jarvishub

Jarvis Hub

Jarvis Hub is the canvas-native creative agent harness that finally keeps your entire design workflow visible and interactive in a single workspace. Instead of forcing you into linear chat windows or complex node-based pipelines, this clever system turns an editable canvas into a shared playground for you and AI agents. It maps out text, storyboards, images, and code as visual nodes that you can easily move, link, and edit. Whether you are generating cinematic storyboards, building interactive web pages, or crafting slide presentations, the agent dynamically inspects the canvas to fix errors and build out new versions while preserving all your feedback. Check it out and bring your creative projects...

🆔 @hackernewsgithubprojects

📦 flocknedevastate/augment-code

Augment Code

Augment Code is the AI assistant that actually understands the unique patterns of massive software projects. Instead of just giving generic coding tips, it reads and indexes your entire workspace to understand how your team specifically writes software. This means when you use its chat interface or inline suggestions in your editor, it tailors every line of code to match your existing setup. It plugs directly into tools like Visual Studio Code and JetBrains, saving developers from manually hunting down context across thousands of files. It makes working in giant, complex projects feel as fast and simple as building a small weekend app.

🆔 @hackernewsgithubprojects

📦 wassgha/rescript

Rescript

Rescript is the open source browser app that finally lets you edit video and audio files just by editing text. You simply drop in an audio or video file, and the app transcribes it fully on your device with word-level timestamps. When you delete a word or phrase from the transcript, the app automatically cuts that corresponding section from the media timeline. It runs completely offline using local intelligence, meaning your media never leaves your computer, and you can even strip out filler words like um and uh with a single click. When you are done, you can export your finished cut right from your browser.

📰 https://news.ycombinator.com/item?id=49065779

🆔 @hackernewsgithubprojects

📦 lzy-dot/specflow

SpecFlow

Vision language models can swallow massive amounts of computing power just processing image tokens, but a clever project called specflow uses spectral heat flow to condense those tokens without losing critical visual data. Instead of randomly dropping pixels or running heavy neural network layers to decide what is important, it models token relationships like heat diffusing across a surface. This elegant physics based approach lets you run visual chatbots like Lava much faster on standard hardware because it keeps only the most essential visual information intact. If you want to speed up visual AI models without destroying their accuracy, this implementation is definitely worth checking out.

🆔 @hackernewsgithubprojects

📦 kimberlyii/omni-prune

Omni-Prune

You can now speed up your multimedia artificial intelligence models by over three times without retraining them at all. This is possible because omni-prune intelligently shrinks the massive mountain of data created when models process audio and video together. Usually, processing these synchronized files takes up massive amounts of computer memory and slows down response times. This project solves that by analyzing both your text prompt and the audio peaks to pinpoint exactly which parts of the video are actually relevant. It then discards the unnecessary background data, slashing memory usage while retaining over ninety-nine percent of the model's original accuracy.

🆔 @hackernewsgithubprojects

📦 zachtong/pyaldic

pyaldic

pyaldic is the python desktop application that finally makes measuring material deformation and strain intuitive and highly precise. Usually, tracking how materials stretch or crack under stress requires clunky, expensive software or running fragile scripts. pyaldic changes the game by wrapping a sophisticated dual-solver engine in a beautiful, multi-language desktop interface. The coolest thing it does is automatic window splitting near boundaries. Instead of getting confused by empty space or background pixels at the edge of a sample, the software dynamically cuts up its tracking zones to only analyze the actual material. You get incredibly smooth, sub-pixel accurate results without writing a single line of code.

🆔 @hackernewsgithubprojects

📦 tencent-hunyuan/twins

Twins by Tencent Hunyuan

Twins is the machine learning framework that uses a clever focal loss strategy to dramatically improve how computers understand and generate high-quality images. Developed by researchers at Tencent Hunyuan, this system is designed to tackle a major challenge in artificial intelligence: unifying the way neural networks represent visual information for both analysis and creation. By combining a powerful dual-head diffusion transformer with a specialized representation autoencoder, it learns to predict unified representations more accurately than standard training methods. It is an incredibly smart setup that simplifies complex multimodal learning, opening up exciting new possibilities for generative media models.

🆔 @hackernewsgithubprojects

📦 worldbench/awesome-embodied-data-pyramid

Awesome Embodied Data Pyramid

Awesome Embodied Data Pyramid is the curated catalog that finally structures the chaotic world of robotics data. Training smart robots to move and interact with physical spaces requires an massive variety of information, which this repository organizes like a pyramid. It maps out everything from real-world robotic trials and human movement videos to virtual simulation environments and web-scale visual databases. It also indexes the advanced models that ingest this data to plan and execute physical actions. This collection gives researchers a structured map of the exact data sources needed to build more capable physical artificial intelligence.

🆔 @hackernewsgithubprojects

📦 anneharrington/studying-cl

studying-cl

studying-cl is the benchmarking library that finally evaluates every major type of continual learning together. While methods like prompting, weight-tuning, and model compression are usually tested in isolation, this project brings them under a single unified harness to see how they perform on the exact same sequential tasks. By running eight distinct methods across four technical families on identical benchmarks like domain shifts and temporal drift, it uncovers when models actually need to learn and when simpler methods work just as well. It is a brilliant, standardized playground for anyone looking to make language models adapt to new data over time without breaking.

🆔 @hackernewsgithubprojects

📦 ewhauser/shuck

Shuck Shell Linter

Shuck scans your shell scripts up to one hundred times faster than legacy tools while packing an entire suite of modern development features. Built from scratch in Rust, this lightning-fast engine acts as a linter, a formatter, and a fully featured language server all in one. It automatically parses multiple dialects like Bash, POSIX shell, and Z-shell to spot security traps, portability bugs, and performance bottlenecks. It even extracts and checks embedded shell blocks hidden inside GitHub Actions workflows. Shuck integrates directly with your favorite editor to give you real-time diagnostics and quick fixes as you type. Grab this tool today to make your shell scripting completely painless.

📰 https://news.ycombinator.com/item?id=47963760

🆔 @hackernewsgithubprojects

📦 kevinmcaleer/snakie

Snakie

Snakie is the physical computing editor that brings your microcontrollers to life with a photorealistic, retro-styled interface. Instead of staring at dry code, this tool visualizes your actual board pinout in real time and lets you drive virtual lab instruments, like an oscilloscope and a multimeter, directly from your running programs. It takes the guesswork out of hardware projects by showing you exactly how your code interacts with the physical world. Whether you are building robots or learning electronics, it is a fun, visual way to master hardware programming. Download it to make your next hardware build much more interactive.

🆔 @hackernewsgithubprojects

📦 allgpt-co/quickvoice

QuickVoice

QuickVoice is the open-source telephony infrastructure that lets you build, run, and self-host your own artificial intelligence voice agents. Instead of paying premium markups to closed third-party services that hide your call logs, this project gives you the complete underlying blueprint to deploy custom voice assistants. You can integrate live telephony, connect your private databases, and trigger outbound calls using an open setup powered by LiveKit. This makes it incredibly easy to automate things like dentist appointment scheduling or customer service desks without letting your sensitive data leave your own servers. Spin up your own secure AI calling system today.

🆔 @hackernewsgithubprojects

📦 nekodificador/comfyui-nkd-vfx-tools

comfyui-nkd-vfx-tools

comfyui-nkd-vfx-tools is the custom node pack that finally gives you precise artistic control over your AI images using actual visual effects tools. Instead of wrestling with text prompts and hoping the model gets your perspective right, you can physically block your shots first. You can place lights on a 3D sphere, find a camera's real focal length, flatten angled walls for easy drawing, and align 3D models directly in your workflow. It translates your visual choices into exact control maps before generating, turning a random lottery into a predictable render with live previews.

🆔 @hackernewsgithubprojects

📦 navikt/cplt

cplt Sandbox

Cplt keeps your sensitive credentials safe by running modern AI coding assistants inside a secure, kernel-level sandbox on your computer. When you let an AI agent write or refactor code in your project, it often needs to run local commands, which can expose your personal private keys, environment variables, and passwords. This tool wraps popular command line tools like GitHub Copilot and Google Gemini inside a strict, deny-by-default environment. It blocks access to sensitive directories and filters outbound network connections, letting the AI read your project files and suggest edits without any risk of stealing your private data.

📰 https://news.ycombinator.com/item?id=48106428

🆔 @hackernewsgithubprojects

📦 gcamilo/management-consulting

Management Consulting

Management consulting is the strategy skill for Claude Code that turns your AI agent into a rigorous, structured thinking partner for high-stakes business decisions. Instead of giving you generic, textbook answers, this tool installs forty-two practitioner-grade frameworks directly into your AI workflow, covering everything from market entry and pricing to organizational design. It enforces strict process discipline, requiring the AI to clarify the decision beforehand, back up every claim with evidence labels like facts or assumptions, and even play devil's advocate against its own recommendations. It is the ultimate way to get clear, visual, and battle-tested strategic analysis right in your terminal.

🆔 @hackernewsgithubprojects

📦 declare-lab/ideagent

IDEAgent

An artificial intelligence system can now generate a high-yield archive of computer science research ideas that are both highly creative and logically rigorous. While typical language models struggle to balance novelty with soundness, a new multi-agent system called ideagent solves this by framing scientific discovery as a quality-diversity search. It manages the evolution of ideas through ancestral lineages, using a multi-objective feedback loop to constantly repair and refine concepts for logical clarity. At the same time, it maintains a lightweight sequential memory to actively compare new suggestions against completed, historical, and even rejected proposals. In comprehensive testing across dozens of computer science topics, this coordinated search strategy outperformed traditional generation...

🆔 @hackernewsgithubprojects

📦 bryanthaboi/gen1recomp

gen1recomp

You can now play Pokemon Red and Blue as native, modern desktop apps without using an emulator. A clever new project called gen1recomp lets you drop in a standard Game Boy ROM, extracts the raw assets, and runs the entire adventure on a hand-written Lua engine. It is not emulation or assembly translation, but a total native rewrite that supports modern controllers, customizable aspect ratios, and even a built-in mod manager. You can zoom out the camera, tilt the overworld, and play with friends over direct network connections. It is the ultimate way to experience Kanto on modern hardware.

🆔 @hackernewsgithubprojects

📦 uditakhourii/adhd

adhd

adhd is the artificial intelligence reasoning tool that prevents coding assistants from getting stuck on their very first thought. Most standard coding tools anchor on the first sensible solution they find, completely missing creative alternatives or hidden architectural traps. This library fixes that problem by taking a single coding prompt and intentionally splitting it across multiple, isolated thinking frames with absolutely no shared context. This forces the artificial intelligence to explore wild and diverse directions, like comparing an economic incentive perspective to a technical fallback approach. A completely separate critic pass then steps in to grade the ideas, filter out bad traps, and deepen the absolute best option. It is...

🆔 @hackernewsgithubprojects

📦 agentscope-ai/qwenpaw

QwenPaw

Run your own private AI companion right from your laptop or cloud setup. With qwenpaw, you get a deeply personalized digital assistant that works across your favorite chat apps and handles real-world tasks like managing calendar jobs or browsing the web. Instead of relying on rigid, closed systems, this project lets you easily expand what your agent can do using custom plugins, custom skills, and even direct browser automation. It bridges the gap between raw language models and a highly responsive, daily assistant. Grab the repository today to deploy your own self-hosted agent and take full control of your digital tasks.

🆔 @hackernewsgithubprojects

📦 nanmicoder/cc-haha

cc-haha

Run your own local-first Claude Code workspace across your desktop and chat apps. With cc-haha, you can manage multiple smart agents, review git code diffs, and deploy specialized skills while keeping your data local and secure. It even lets you interface with Claude directly through platforms like WeChat, Telegram, and WhatsApp, or let a virtual desktop companion monitor your tasks. It is the ultimate bridge for developers who want a powerful, local-first artificial intelligence assistant that lives right in their existing messaging workflows and development environments. Check it out and bring your developer workspace into the future.

🆔 @hackernewsgithubprojects