📦 moeblack/moss-image-prompt-skill

Moss Image Prompt Skill: The Ultimate Prompt Compiler

Moss Image Prompt Skill is the prompt compiler that turns vague ideas into cinematic, photorealistic AI art. Most people waste hours guessing at text boxes, but this open-source rulebook acts like a strict editor that organizes your words into a precise recipe. It solves the problem of messy generations by providing a strict formula for lighting, camera angles, and character details that actually works with the Moss Image model. The real magic is its ability to replicate movie stills without using movie names, relying purely on technical descriptions of light and film grain to capture the exact mood.

🆔 @hackernewsgithubprojects

📦 lucapazienza/gravityrush_pc

Play Gravity Rush on PC Without an Emulator

Run Gravity Rush Remastered natively on your PC without an emulator. This community project executes the original PlayStation four code directly on your processor using a custom runtime that mimics the console system. It swaps the console graphics engine for Vulkan, enabling native forty kilohertz rendering, high quality shadows, and support for Nvidia DLS. The DualSense gyro controls the gravity camera exactly like the handheld. You get full save support, streamed music, and a dark themed launcher with built in mod support. The opening city scenes run smoothly with working audio and trophies.

🆔 @hackernewsgithubprojects

📦 tunr-dev/tunr

tunr: Safe AI Data Access

tunr lets AI agents peek into your company's private databases without breaking security rules. The problem is that AI tools often need access to sensitive internal data, but companies don't trust external agents with their passwords or permissions. tunr solves this by creating a secure bridge where the customer approves every single request for a specific time and purpose. The coolest part is that the agent can only read the data, never write to it, and the connection is encrypted all the way to the customer's server. Once you have this setup, adding new agents is as simple as one approval.

📰 https://news.ycombinator.com/item?id=50086671

🆔 @hackernewsgithubprojects

📦 worldplay2/worldplay2

WorldPlay2: Playable AI Worlds

WorldPlay2 generates interactive video worlds where you control the movement and the story as it happens. Instead of watching a pre-rendered clip, you type a prompt describing a scene and then steer a character through it in real time using your keyboard. The model builds the video frame by frame, responding instantly to your inputs so you can walk around a forest or change the weather on the fly. This solves the problem of static video generation by creating a living environment that reacts to you. It is a fascinating step toward fully playable digital spaces powered by artificial intelligence.

🆔 @hackernewsgithubprojects

📦 yiboz2001/ultratex

UltraTex: 2K 3D Texturing, 90x Faster

UltraTex just made high-resolution 3D texturing dramatically faster by teaching AI models to ignore what they do not need to see. Traditional systems waste massive amounts of processing power analyzing empty background space, but this new tool cuts that waste by over ninety percent. It works by letting the algorithm skip over blank areas and only pay attention to the actual object in front of it. This means you can generate incredibly detailed two thousand by two thousand pixel textures for 3D models without waiting for hours or burning through your entire compute budget. It is a clever way to make heavy graphics generation practical for everyday artists and developers.

🆔 @hackernewsgithubprojects

📦 amitshekhariitbhu/ai-agents-tutorial

AI Agents Tutorial: From Scratch to Multi-Agent

AI Agents Tutorial is the step-by-step guide that demystifies how artificial intelligence actually thinks and acts. Instead of just throwing code at you, it walks you through the real mechanics behind the magic. The coolest part is how it breaks down the agent loop, showing you exactly how a machine decides what to do next, like watching a detective solve a case in real time. It covers everything from basic tools to complex systems where multiple bots work together. If you want to build AI that truly works, this is your roadmap. Stop guessing and start understanding how these systems actually function.

🆔 @hackernewsgithubprojects

📦 iamwavecut/ninfer-all

Run Massive AI Locally on Your GPU with Ninfer All

Run massive AI models directly on your graphics card without touching the cloud. Ninfer All takes the messy, fragmented world of local inference and consolidates it into one polished engine for top-tier consumer GPUs. Instead of juggling different forks for every feature, it bundles everything you need, from specialized weight formats to advanced routing profiles, into a single, coherent tool. This means you get serious, high-performance AI capabilities running entirely on your own hardware, giving you total control over your data and speed. It is the definitive way to run powerful, next-generation language models locally, turning your personal computer into a serious AI workstation without the headache of manual integration.

🆔 @hackernewsgithubprojects

📦 nahumlitvin/prismantis

Prismantis Makes Your Terminal Charts Pop

Transform plain text replies into vibrant, interactive visuals inside your terminal. Prismantis takes the raw data from your coding assistant and draws colorful tables, flowcharts, and bar graphs directly in your command line. It solves the problem of staring at wall-of-text replies by using syntax highlighting and custom color themes to make complex information instantly readable. You can even click a tiny button to copy any chart or code block straight to your clipboard. This turns a static terminal into a dynamic dashboard without leaving your workflow. If you code in a terminal, this simple plugin makes your daily work far more visual and organized.

🆔 @hackernewsgithubprojects

📦 5uck1ess/cicero

Cicero: The Voice Interface for Your Coding Agents

Cicero turns your coding agents into a spoken, real-time assistant. It sits next to your favorite AI coder, letting you talk to it from any browser or even a phone call. The coolest part is that you can interrupt it mid-sentence just like a real conversation. It listens to your tone, understands when you finish, and actually does the work in the background. If you ask it to fix a bug, it handles the coding and opens the pull request for you. Then it calls you back when the job is done. All of this runs locally, so your voice data stays on your machine.

🆔 @hackernewsgithubprojects

📦 7api/uncensored-ai-api

One Key for Every AI Model

Uncensored AI API is the developer gateway that finally puts every major AI model behind a single key. It solves the headache of juggling separate accounts and bills for Claude, GPT, and Gemini by routing them all through one compatible endpoint. The surprising part is that it includes uncensored variants, meaning you can access versions with fewer built-in refusals for research or creative work. You just change your base URL and API key to start using it with your existing tools. It is pay-per-token with no subscription, so you only pay for what you use.

🆔 @hackernewsgithubprojects

📦 actionow-ai/vvv

Turn Photos Into Films With VVV

Turn a folder of photos into a finished, music-synced wedding film by simply asking your coding agent. The open source project called Vibe Vision Video lets you build a video template once and reuse it across different agents, saving time and effort. The most surprising feature is its recreate mode, where you give the agent a reference video you love. It studies the rhythm and style, then rebuilds the scene with your own photos and text. The agent handles the heavy lifting of layout, timing, and rendering checks. You just watch the sample and request changes. This makes professional video production accessible to anyone, not just editors.

🆔 @hackernewsgithubprojects

📦 helloianneo/ian-xhs-carousel

Ian XHS Carousel: The AI Skill for Perfect Xiaohongshu Carousels

Ian XHS Carousel is the AI skill that turns messy articles into polished Xiaohongshu carousels without design work. You simply paste your text, and it drafts a storyboard for you to approve before generating a set of portrait images. The standout feature is its collection of twenty distinct templates, which ensures every page looks like a professional product interface rather than a generic graphic. It handles both light and dark themes, so your content stays consistent no matter the background. This is ideal for creators who want clean, readable visuals without hiring a designer.

🆔 @hackernewsgithubprojects

📦 qianshuixia/shrimpreel

ShrimpReel: Turn One Photo Into A Viral AI Video

ShrimpReel lets you spin up a complete, standalone website where users upload a single photo and receive an AI-generated video that mimics a specific viral internet trend. The platform solves the tedious task of building these niche tools from scratch by providing a ready-made template that handles user sign-in, credit payments, and video generation all on a single cloud platform. Instead of coding each feature manually, you simply clone the project, swap out the text and media for a new keyword, and deploy a fully functional site. This approach means you can launch a unique web page for every trending dance or filter without worrying about the underlying infrastructure.

🆔 @hackernewsgithubprojects

📦 komi-store/komi-store

Komi Store: The App Store for GitHub

Komi Store treats code repositories on GitHub, Codeberg, and Forgejo as if they were standard software stores. Most people view code hosting sites as places to read text and track bugs, not to find and install finished programs. This application bridges that gap by scanning public release files and presenting them in a familiar store interface. It works on Android, Windows, macOS, and Linux, letting you browse, discover, and install these open-source tools with a single click. The project is built with a single codebase to ensure the experience feels consistent everywhere you run it.

🆔 @hackernewsgithubprojects

📦 mrdoob/slop-newyork

Slop New York: The Open-Source City Simulator

Slop New York renders a navigable, three-dimensional map of Manhattan directly inside your web browser. Instead of requiring a massive download or a powerful gaming console, this open-source project uses modern graphics standards to let you explore the island from the street level up to the sky. The most striking aspect is the freedom of movement: you are not limited to walking, but can seamlessly switch between driving a yellow cab, riding a bicycle, or piloting a helicopter. This is not a finished video game, but a flexible creative tool designed for developers and artists to remix, break, or turn into their own visual experiments.

🆔 @hackernewsgithubprojects

📦 chipfighter/context-attachment-manager

The AI Image Chat Fix

Keep long image chats alive in AI tools like Codex and Claude by letting the model manage its own attachments. The context attachment manager solves the lag problem where every message resends the entire conversation history, causing massive slowdowns. The surprising part is that the AI quietly writes notes about every image you share, describing what it sees and what you said about it. This means you can search for a picture by remembering a vague detail like the best face or a rainy night, even after the chat has been compacted.

🆔 @hackernewsgithubprojects

📦 mzhao-ouo/touchable

Touchable: Play Switch Games Natively on iPhone

Touchable is the gaming tool that finally makes ported Switch titles feel genuinely native on your iPhone. If you have ever played a console game on your phone using clunky emulator overlays, you know the struggle. This project takes the raw game files you own and rebuilds them into a true Apple app. It replaces those awkward screen overlays with beautiful, physical-feeling touch buttons that look right at home in your pocket. You run the build on your Mac, install the result on your phone, and play. No more fighting with fake controls.

🆔 @hackernewsgithubprojects

📦 lnpandy/nx2apple

Play Switch Games on iPhone

Run your own Nintendo Switch games directly on an iPhone or iPad. This project, called N X Two Apple, solves the biggest technical hurdle for large titles like Tears of the Kingdom, which contains over three hundred thousand tiny files. Standard installation tools crash when trying to handle that many files, and copying them to your phone takes hours. This tool fixes that by packing all the game data into a single, compact archive file. It rewrites the game code to work on Apple chips and bundles everything into a slim package that installs in minutes instead of hours.

🆔 @hackernewsgithubprojects

📦 op7418/guizang-tv

Guizang TV: The Calm Home Screen for Xiaomi TVs

A calm, readable home screen now exists specifically for Xiaomi smart TVs. Guizang TV replaces the cluttered default interface with a large text list and rotating wallpapers, solving the problem of small, unreadable icons on living room screens. It also adds direct HDMI switching, letting you jump straight to game consoles or media boxes without navigating menus. This is interesting because it turns a standard smart TV into a focused media hub, prioritizing clarity over visual noise. If you own a Xiaomi TV, this project is your path to a cleaner, more usable experience right from the start.

🆔 @hackernewsgithubprojects

📦 mclhm8182/tvhome-zh

tvhome-zh: The Apple TV Home Screen for Android

Your living room TV suddenly looks and feels like an Apple TV, right on the screen. This project, known as tvhome-zh, swaps out the cluttered, ad-filled default interface for a sleek, black home screen with a glowing dock. The most surprising part is that it arrives as a tiny, ready-made file you can install instantly without compiling any code. It solves the problem of technical complexity by turning every setting into a simple button on your remote control. You can pick your wallpaper, choose your apps, and adjust the layout without touching a single computer terminal. It is a clean, lightweight replacement that respects your time.

🆔 @hackernewsgithubprojects

📦 img2threejs/plugin-interior

Turn One Photo Into a Walkable 3D Room

Take a single photo of a room and turn it into a fully walkable three-dimensional scene you can edit and explore. This project, called plugin interior, uses artificial intelligence agents to build a real three-dimensional model from a flat image, creating an interactive viewer you can actually navigate in a browser. It is not just a filter or a simple picture; it generates an editable scene where you can adjust furniture, change materials, and look around from different angles. The coolest part is that it handles the hard technical lifting for you, turning a static snapshot into a living digital space.

🆔 @hackernewsgithubprojects