298 subscribers
14K videos
14.6K links
Download Telegram
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ realcodebear/saltline-fable-5.2

Saltline Fable 5.2: The One-Prompt FPS

Saltline Fable 5.2 is the browser-based first-person shooter that proves a single prompt can build a fully playable game. It is a rainy container port level where you hunt three uplink relays while dodging drone and robot dog enemies. The whole experience lives in one file, with no external assets to download. It was generated by an AI model in one continuous run, meaning no human touched the code after the initial idea. You can run it locally by serving the file in your browser and clicking to start. It is a fascinating look at what modern AI can do when given a clear creative goal.

πŸ†” @hackernewsgithubprojects
Media is too big
VIEW IN TELEGRAM
πŸ“¦ scottduncanai/pid-to-knowledge-graph

Turn Plant Diagrams into AI Knowledge

pid-to-knowledge-graph lets you turn complex factory diagrams into a smart map that artificial intelligence can actually understand. It takes flat P and I D drawings and builds a living network of pipes, sensors, and controls. The coolest part is how it handles scanned paper drawings. It sharpens faded lines and uses text recognition to find tiny tags that would normally be invisible. Once built, you can ask plain English questions like which valve controls this tank. The graph answers by tracing the logic straight from the original drawing. It turns static engineering sheets into interactive data you can query instantly.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ rivet-dev/open-orange

Run Cloudflare Workers anywhere with OpenOrange

OpenOrange is the open source tool that lets you run your existing Cloudflare Workers code on AWS, Google Cloud, Azure, or your own machines without changing a single line of your application. The real surprise here is how it handles data: instead of leaving you stranded, it moves your Durable Objects off Cloudflare to new servers while your app keeps serving traffic, meaning you can switch providers without a single second of downtime. This gives you the exact same developer experience you are used to, but with a bill based on raw machine costs rather than complex platform metering.

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

SnowLuma: The QQ Bridge for Developers

SnowLuma is the TypeScript runtime that makes your existing QQ client work with modern app protocols. Instead of writing complex code to talk to Tencent's servers, it intercepts your live session and converts every message and event into a standard format your apps can understand. This means you can build bots or manage multiple accounts through a simple web dashboard or a developer SDK without touching raw data. It is the clever way to connect legacy chat systems to new tools, keeping everything organized and accessible in one place. If you want to build with QQ data without the headache, this is the bridge you need.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ yetone/rewind-for-x

Rewind for X

Recall the tweet you missed before your feed refreshed. Rewind for X is a Chrome extension that saves every post you scroll past on your home timeline, right on your device. It captures the text, images, and authors so nothing disappears when the app updates. The coolest part? It reads the words inside your pictures, letting you search for specific phrases even if the caption is blank. Since everything stays local, your privacy is completely safe. No more hunting for that perfect recipe or funny quote you saw twenty minutes ago. Just hit a shortcut, find it instantly, and get back to browsing without losing your place.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ gloriaameng/awesome-agent-harness

Awesome Agent Harness: The Map for AI Agents

Awesome Agent Harness is the visual map that finally makes sense of all the different ways you can build large language model agents. Most people just guess which tools to use, but this project organizes thirty-four real systems into a clear grid so you can see exactly how they handle tasks and what evidence supports their design. It cuts through the noise by letting you search and filter the landscape instantly. If you are building agents, this is the cheat sheet you need.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ neywa-labs/mlx-audio-swift

MLX Audio Swift: On-Device Audio AI for Swift Developers

The modular Swift SDK that runs powerful audio AI directly on your Mac. MLX Audio Swift is a developer toolkit that lets you build apps with real speech intelligence without sending data to the cloud. It handles everything from turning text into natural voices to transcribing recordings, all using the built-in chip on Apple devices. This means your apps stay private and fast because the heavy math happens locally. Instead of relying on external servers, you get immediate results right on your machine. It is a clean way to add smart audio features to your projects.

πŸ†” @hackernewsgithubprojects
Media is too big
VIEW IN TELEGRAM
πŸ“¦ threerocks/article2media

Turn Any Video Into Viral Content

One link turns a long article or a raw video into polished, ready-to-post visuals. Article Two Media is an open-source skill that handles the heavy lifting. You paste a link, and it automatically transcribes the video, rewrites the text, and designs professional layouts for social media. It creates phone-sized long images, clean card decks, and even vertical videos with AI voiceovers. The coolest part? It runs entirely on your computer. No cloud servers, no hidden costs, and zero external dependencies. It just works. Stop spending hours on manual design. Let your AI assistant do the formatting while you focus on the next idea.

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

Maxagent gives AI coding agents a permanent memory

Maxagent installs a persistent memory system into the AI coding tools you already run, letting them remember exactly which fixes worked and which commands fail. Instead of guessing through the same errors every single day, the agent pulls up a short proven recipe the moment it hits a snag. This small project turns your daily work into a reusable library that makes every task faster and more reliable. You get a tool that stops your computer from repeating the same mistakes, letting you focus on building rather than debugging. The result is a coding assistant that actually gets smarter the more you use it.

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

ZigZag: Let AI Stitch Your Logos

Turn flat vector art into actual machine embroidery with ZigZag, a free desktop tool that handles the messy math of stitch planning for you. You drop in an image, and the software automatically calculates exactly how the needle moves to fill shapes and stitch outlines without breaking thread. The coolest part is built right into the interface: you can talk to an AI agent that actually builds the design alongside you. It picks colors, adjusts the stitch patterns, and places objects on the canvas in real time. It is not just a passive converter.

πŸ“° https://news.ycombinator.com/item?id=50470488

πŸ†” @hackernewsgithubprojects
Media is too big
VIEW IN TELEGRAM
πŸ“¦ wy51ai/easycolab

EasyColab: Your Cloud GPU in Your Pocket

EasyColab is the desktop studio that turns your phone into a remote control for massive cloud graphics cards. Most people think using heavy video and image tools requires a room full of expensive servers, but this app lets you wake up a powerful computer in the cloud with a single tap on your phone. It automatically spins up a machine, runs the program, and sends the finished video right back to your gallery. The best part is that it shuts the computer down after thirty minutes of inactivity, so you stop paying the moment you stop working. It is like having a private, on-demand supercomputer that lives in your pocket.

πŸ†” @hackernewsgithubprojects
Media is too big
VIEW IN TELEGRAM
πŸ“¦ xiaoerofficial021-collab/flowhmr-windows

FlowHMR Windows: Turn Video into 3D Animation

FlowHMR Windows is the desktop app that turns ordinary video footage into exportable 3D character animations. You simply feed it a video clip of a person moving, and the software reconstructs a full 3D body model that perfectly mirrors those movements. It solves the problem of needing expensive motion capture hardware by letting your computer do the heavy lifting. The real magic is exporting those moves directly into FBX files, which you can then drop into game engines or animation software. This makes professional-grade motion capture accessible to anyone with a decent Windows PC.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ t4t5/owly

Owly: AirDrop for Linux

Your Linux laptop can now receive files directly from an iPhone using the same hidden Wi-Fi link that Apple uses for AirDrop. A project called Owly acts as a bridge, taking over your Wi-Fi card to create a direct radio connection with your phone. It works on standard Linux hardware, letting you send and receive photos just like you would between two iPhones. It solves the problem of needing to email yourself files or use third-party apps to move data across ecosystems. The software handles the tricky part of the radio protocol so you can just drag and drop.

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

Solana Vigil: The 60-Second Validator Check

Vigil runs entirely on your server without ever asking for your private keys, ensuring it can never steal funds or break your validator. Running a Solana node usually means gluing together a dozen different scripts to track votes, disk space, and system health, but Vigil bundles all of that into a single, tiny program. It connects to your local node to read its status, spotting problems like low storage or missed blocks before they cause downtime. Because it is strictly read-only and built from one static file, it is safe to run permanently without worrying about hidden costs or security risks.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ setyamickala/fastmint-fcfs-agentic

Fastmint FCFS Agentic: 2-Second Mint Precision

An AI agent just landed a free NFT mint in a block two seconds after the sale opened, and it never touched a private key. Fastmint FCFS Agentic is a skill that teaches coding tools like Claude Code to fire a single transaction at the exact millisecond a sale starts. It solves a real problem: standard wallets are too slow for first come, first served drops, often missing out entirely. By using Binance’s secure backend to handle the actual signing, the agent only focuses on timing the shot perfectly.

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

Gester: The iOS Sandbox Escape Tool

Gester is the iOS tool that turns a standard app container into a key for the system's hidden control panel. It solves a specific problem: normal apps are locked in a sandbox, unable to touch critical system files. Gester finds a clever gap in how the operating system validates paths, allowing it to safely slip outside that box just enough to edit the device's core configuration files. This means you can flip on hidden developer features, change how the screen behaves, or customize the system wallpaper in ways Apple never intended for public use.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ danyuchn/claude-work-mode

Claude Work Mode: Stop the AI Interruptions

Claude Work Mode is the session tool that finally stops Claude Code from asking you permission for every little step. It fixes the annoying habit where the AI pauses to confirm reversible actions like local file edits or simple commits, which really kills your focus. By switching to a specific work mode with one simple command, you tell the model to just handle the routine stuff and only flag the truly critical decisions at the end. This keeps your main chat clean and saves your expensive usage quota for the tasks that actually need your attention.

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

xj-scanner: Let an AI Agent Scan Your Jeep

xj-scanner is the hobby tool that lets an AI agent talk directly to a 1994 Jeep engine computer. Most people think old cars are dumb, but this project proves you can plug a three-dollar microcontroller into the diagnostic port under the hood and wire it to your laptop. The real magic is that it hands the wheel to a coding agent. You don't write code or guess at wiring. You just tell the assistant to help you connect the car. It walks you through finding the right pins, reading live engine data, and even running built-in tests.

πŸ†” @hackernewsgithubprojects
This media is not supported in your browser
VIEW IN TELEGRAM
πŸ“¦ likely7/veyra-nrvideo

Veyra NRVideo: The Secret to 120 FPS Gaming on PC

Stop watching your games look like a blurry dream while your console runs at a silky smooth sixty frames per second. Veyra NRVideo is a Windows player that uses your graphics card to magically add frames and sharpen pixels in real time. It’s not just a viewer; it’s an enhancement engine that takes a standard video file or a live capture from your console and turns it into a crisp, ultra-smooth experience. Think of it as giving your eyes a cheat code, letting you see every detail in a high definition, high frame rate format that usually requires expensive new hardware.

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

Avenoxbeyin: Your AI Second Brain

Avenoxbeyin is the local-first second brain that finally lets an AI agent remember things without sending your private notes to a distant server. It plugs directly into your existing Obsidian vault, turning a pile of Markdown files into a persistent memory system that works across different chat sessions. You simply tell it to set up, and it handles the rest, ensuring your data stays on your machine and under your control. It is a clever way to give coding tools a long-term recall capability without the usual cloud privacy headaches.

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

Grokbot Clone: Your Own AI Office

Grokbot Clone is the self-hosted AI team manager that gives every bot its own physical workspace. You can hire digital coworkers who actually sit at a computer, browse the web, run code, and finish jobs while you watch their screens. The coolest part is that these bots can talk to each other, share files, and even call you on the phone. It turns your local machine into a headquarters where you direct a team of tireless assistants. If you have ever wanted a software team that never sleeps and reports directly to you, this is the blueprint for building it yourself.

πŸ†” @hackernewsgithubprojects