📦 theguysudo/enzo

Enzo: Your Private AI Workspace

Enzo is the self-hosted AI workspace that puts total control back in your hands. It runs entirely on your own infrastructure, letting you plug in your own API keys from providers like Groq or Google so you pay the model directly without any middleman markup or subscription fees. The standout feature is its agent builder, where you describe a task in plain English and the system drafts a complete operating manual for the agent, which then learns from your activity to improve over time. It also gives you access to over three hundred models through one simple catalog. No accounts, no hidden meters, just your keys and your data.

🆔 @hackernewsgithubprojects

📦 addyosmani/clarity

Clarity: Write Like You Talk

Take a coding agent and teach it to write like a human, not a robot. That is exactly what Clarity does. It is a skill that plugs into your development tools to help you draft, rewrite, or review prose with genuine clarity. The coolest part is interview mode. Instead of guessing what you mean, the agent asks you specific questions about your experience, your audience, and your argument. It then builds the piece using your actual words and examples, rather than inventing generic filler. It feels less like using an AI and more like having a brilliant editor who actually listens.

🆔 @hackernewsgithubprojects

📦 kivutar/goro

Goro

This entire MMORPG client is written in Go and runs on your GPU without a single C header. Goro recreates the classic two-thousand-eight Ragnarok Online experience using a modern Vulkan pipeline. It handles everything from character creation to complex guild management and pet hatching. Because it is pure Go, it compiles easily on any platform without messy external dependencies. It even includes tools to pack and unpack game resources. If you want a clean, hackable codebase that actually runs the game, this project is your new favorite weekend experiment.

📰 https://news.ycombinator.com/item?id=49767075

🆔 @hackernewsgithubprojects

📦 matheusht/redthread

RedThread: The AI Bodyguard

RedThread attacks your AI models to find the exact prompts that break them, then automatically builds and tests the fix. It doesn't just guess at security holes; it runs a full loop of generating attacks, scoring the damage, and creating specific guardrails to stop those exact exploits. The coolest part is that it finds the smallest possible trigger for a jailbreak, so you aren't guessing what went wrong. It proves the defense works before you deploy it, turning guesswork into solid proof. If you ship AI, this is how you keep it safe without the guesswork.

🆔 @hackernewsgithubprojects

📦 malerba118/element-mirror

Element Mirror: Real-Time Canvas Reflections

Element Mirror lets you project any live website element onto a canvas in real time, capturing everything up to sixty times a second. This is particularly cool for creating drag ghosts or blurred video backdrops without duplicating your code. Instead of cloning the DOM, it uses a shared capture system that records the actual pixels, shadows, and animations. This means a single source can feed multiple delayed mirrors, allowing you to build smooth motion trails that cost almost nothing extra. It is a clever way to keep visual effects perfectly in sync with your interface. The best part is that it handles the heavy lifting for you.

🆔 @hackernewsgithubprojects

📦 logan-markewich/jeff

Jeff: The Self-Hosted Alternative for TypeSafe's Jev

Jeff is the self-hosted server that lets you run TypeSafe's classification API on your own hardware using a smaller, open-source model. Instead of sending sensitive customer data to a third-party cloud for analysis, this project runs locally on a GPU to answer specific questions about text, such as determining tone or urgency. It acts as a drop-in replacement, meaning you can point existing applications at it without rewriting code, which is particularly useful for teams needing strict data privacy or lower costs.

🆔 @hackernewsgithubprojects

📦 mrnugget/jev-shell-history

Jev Shell History: AI-powered terminal autocomplete

Jev Shell History is the terminal plugin that uses artificial intelligence to guess which command you are about to type. Unlike standard autocomplete tools that only match the first few letters you enter, this tool looks at your entire recent command history to understand your intent. It works by asking a specialized model to pick the most likely command from your last hundred entries based on what you have already typed. The suggestion appears in grey next to your cursor, and you can accept it with a single keystroke. This is perfect for developers who use complex commands frequently, as it remembers your specific workflows and shortcuts.

🆔 @hackernewsgithubprojects

📦 cvxv666/fomo-robinhood-radar

FOMO Robinhood Radar

The most surprising fact about this tool is that it can detect when a token transfer is actually fake. FOMO Robinhood Radar tracks specific traders on Robinhood Chain and verifies every single trade by checking the blockchain receipts. It solves a common problem where data feeds mistakenly credit wallets for tokens they never bought. This matters because it filters out noise to show you which traders are genuinely making good moves. It runs for free and gives you clear, verified signals instead of confusing data. In short, this is a free tool that proves exactly who is buying what, so you can trust the data before you act on it.

🆔 @hackernewsgithubprojects

📦 ni-co-la-s/gemmeh

Gemmeh: Building an AI from Zero

Build a language model from absolute scratch. This project walks through every single step, starting with raw web data and ending with a chatbot you can actually talk to. The coolest part? They trained it only on data from before twenty twenty-four. This intentional knowledge cutoff lets researchers measure exactly how surprised the model gets when it encounters recent events. It is like giving an AI a time machine to see the future. You get the full stack, from making the word processor to final training. It is a brilliant, transparent way to learn how these models really work. Check out the code to see the whole journey.

🆔 @hackernewsgithubprojects

📦 wangtong627/artemis

Artemis: AI That Debates to Find Tumors in Video

Artemis turns rough, hand-drawn scribbles on medical video frames into precise, reliable maps of intestinal tumors, all without needing hours of perfect manual labeling. The surprising part is how it works: it doesn't just guess. Instead, a vision-language agent acts like a committee, debating which initial guesses are actually trustworthy before spreading those confident predictions forward and backward through the footage. This debate-and-judge approach filters out noise, meaning doctors get accurate diagnostic tools from minimal effort. If you want to know how AI can cut through the hassle of medical data annotation, this project shows exactly how to do it.

🆔 @hackernewsgithubprojects

📦 fredy-zhang/prdit

PRDiT: The 3D CT Generator

PRDiT synthesizes full, high-resolution 3D CT scans directly from noise, skipping the usual blurry downsampling traps that plague other medical image generators. The real trick here is its two-step approach: it first nails down the big, low-frequency shapes of the lung or chest using a simple local estimator, and only then uses a smart transformer to add in the fine, high-frequency details across the entire volume. This coarse-to-fine strategy makes training way more stable and keeps those tiny, critical structures sharp without losing quality. It beats current state-of-the-art models on major datasets like LIDC and RAD-ChestCT, proving that you can generate realistic, pixel-perfect medical volumes without sacrificing speed.

🆔 @hackernewsgithubprojects

📦 celestoai/celesto

Celesto: The AI Agent Computer

Celesto is the virtual computer infrastructure that gives AI agents a safe, persistent place to actually work. Instead of just chatting, the agent gets its own isolated machine where it can open a browser, run code, and keep files between sessions. The coolest part is that these computers are built on micro virtual machines, so you can snap a photo of the entire system state and restore it later. This means your agent never forgets its progress. If you are building tools where an AI needs to handle real tasks, this is how you give it a brain without giving it the keys to your house.

🆔 @hackernewsgithubprojects

📦 kalldrexx/emacs-zero-to-ide-journey

Building An IDE From Scratch In Emacs

Transform a bare text editor into a full development environment by writing actual code directly inside the application. This journey documents the step by step process of turning vanilla Emacs into a daily driver that replaces heavy graphical tools. The most surprising part is that you do not use a rigid settings file or a plugin menu. Instead, you type live code that changes how the editor behaves immediately. This approach lets you understand exactly why every feature works, rather than just trusting a prebuilt package. It is a deep dive into the mechanics of building your own tools from the ground up.

🆔 @hackernewsgithubprojects

📦 tajirax/cottendns

CottenDNS: Bypassing Censorship Through DNS

CottenDNS is a censorship circumvention tool that hides internet traffic inside standard DNS queries, specifically designed for countries where deep packet inspection blocks normal VPNs. By tunneling data through UDP port fifty-three, it slips past firewalls that whitelist DNS as a trusted channel. The project prioritizes speed and resilience on slow, lossy connections by using aggressive compression and custom error recovery, ensuring every byte counts. It includes preconfigured profiles for different network conditions, allowing users to tune their setup for maximum stability without complex manual adjustments. This tool proves that even when most routes are blocked, a creative use of a fundamental internet protocol can keep you connected.

🆔 @hackernewsgithubprojects

📦 inlitx/gymmane

GymMane: The Gym Tracker That Hates Your Internet

GymMane is an Android fitness tracker that physically cannot connect to the internet. It does not request network access, meaning your workout data stays entirely on your phone. You tap muscles on a body map, log your sets, and watch your personal records climb without any account, ads, or cloud servers. It is a private, offline tool for tracking your gym progress. If you want your training data to remain yours, this app is the way to go.

🆔 @hackernewsgithubprojects

📦 iterative-computer/powermove

Powermove: The Video Editor That Rewrites Itself

Powermove lets you ask a computer program to build new tools for your video editing session. Instead of sticking to a fixed menu, the app writes fresh code to add effects or panels right in front of you. If the result looks wrong, you can correct it or undo the change completely. This matters because it turns a static tool into a flexible workspace that adapts to your specific creative needs. You get a video editor that grows and changes based on your requests, making the creative process faster and more responsive to your workflow.

🆔 @hackernewsgithubprojects

📦 kshetrajna12/reflex

Reflex: The Tiny AI That Knows Its Limits

Reflex is a tiny AI model that answers questions by giving you honest percentages instead of written text. Instead of generating paragraphs that might be wrong, it reads your data once and instantly tells you exactly how sure it is about each possible answer. If it says ninety percent, it is actually right ninety percent of the time. This is a huge deal for tasks like sorting support tickets or checking photos, because you can set strict rules on the numbers without worrying about confusing the machine. It runs fast on your own computer and keeps your data private.

🆔 @hackernewsgithubprojects

📦 mr-september/ninfer-2080ti-22g

ninfer-2080ti-22g: Run 35B Models on a 2080 Ti

ninfer-2080ti-22g runs a thirty-five billion parameter Qwen model on an eight gigabyte 2080 Ti by squeezing data into four-bit precision. It replaces the usual memory-heavy inference stack with custom C++ kernels that map directly to the card's hardware. This allows heavy local generation without cloud costs or massive power bills. The project proves that smart quantization and low-level optimization can push consumer graphics cards far beyond their nominal limits. It is a practical toolkit for developers who want serious local performance on budget hardware. Your takeaway is that modern GPUs can handle massive language models if you write the code to match the silicon.

🆔 @hackernewsgithubprojects

📦 atoonk/packetio

Packetio: Write Fast Network Code in Go

Stop wrestling with low-level network drivers. Packetio lets you build high-speed packet senders in Go without drowning in complex configuration. It unifies four different hardware access methods under one simple interface. You write your loop once, then just swap the import line to match your specific machine. Whether you are on a high-end server or a basic laptop, the code stays the same. This kills the setup headache and lets you focus on logic. It is the fastest way to get real data moving on the wire today. Give it a try and save yourself hours of debugging.

📰 https://news.ycombinator.com/item?id=49607200

🆔 @hackernewsgithubprojects

📦 chinmay-sawant/gowkhtmltopdf

gowkhtmltopdf: Go's Browser-Free PDF Engine

gowkhtmltopdf is the pure-Go template engine that converts HTML directly into PDFs without needing a browser or external binaries. It solves the messy dependency headache of tools that rely on heavy system libraries to render documents. You simply feed it an HTML string with styling, and it processes the layout internally to spit out a clean, professional PDF or image. It is perfect for generating invoices, certificates, and reports straight from your Go application code. The best part is that it runs everywhere without the usual setup nightmares, giving you full control over your document generation pipeline.

🆔 @hackernewsgithubprojects