📦 mrhlaingbwardev/agentic-hacking-skills

Agentic Hacking Skills

Agentic Hacking Skills is the open-source toolkit that turns a standard coding assistant into a capable web security researcher. This repository contains twenty-eight specialized instruction files and a custom server connection that lets the AI directly operate professional testing tools like Nuclei and Nmap inside a Linux environment. Instead of just reading documentation, the AI can actively scan for vulnerabilities, analyze network traffic, and execute complex reconnaissance tasks. It solves the problem of needing a separate terminal for every manual step by allowing natural language commands to drive the entire security workflow.

🆔 @hackernewsgithubprojects

📦 cis-c0/rfsentinel

RFSentinel: Detect Police Tech With Your Android

Detect hidden police surveillance with your Android phone using RFSentinel. This free open source app spots body cams license plate readers and drones by listening to radio signals. It identifies fake cell towers and smart glasses without needing root access. You get instant alerts on a live map. It turns your phone into a radar for law enforcement tech. The best part is it works on any Android device. No special hardware required. It uses built in bluetooth and wifi. You can track your location and avoid surveillance. The interface is simple and easy to use. It is perfect for privacy conscious drivers. You know exactly what is watching you.

🆔 @hackernewsgithubprojects

📦 huacnlee/noop

Noop

A single program quietly runs your entire digital workforce from your own computer, not a distant server. It manages a private team of artificial intelligence agents that break down complex goals and execute them in parallel. The system learns your specific decision-making patterns over time, allowing it to handle routine tasks autonomously while reserving your attention for critical choices only. When a decision is needed, you receive a simple notification with a few options, answering with a single tap. This setup keeps your data entirely on your own hardware, eliminating the noise of raw errors and unnecessary interruptions.

🆔 @hackernewsgithubprojects

📦 flitai-io/gpuworks

GPUWorks: Run Your Own AI Cloud in 40MB

GPUWorks is the ultra-lightweight platform that turns any mix of hardware into one unified AI cluster. It solves the headache of juggling different graphics chips by letting you manage Nvidia, AMD, and even Apple Silicon nodes from a single, tiny download. You just plug in your own servers or rented cloud instances, pick a model, and it handles the rest, serving your AI behind one simple address. It is the perfect way to take control of your own compute without the bloat. If you want to build your own AI stack, this is the tool to watch.

📰 https://news.ycombinator.com/item?id=50367630

🆔 @hackernewsgithubprojects

📦 sudo-py-dev/fmx

The media converter that does it all

FMX is the all-in-one media converter that turns your messy file pile into exactly what you need without the usual headache. Built in Rust for speed, this lightweight desktop app bundles the heavy lifting tools right inside, so you never have to hunt for missing dependencies or install extra software. The coolest part? It handles video, audio, images, and even PDFs in a single, clean queue, letting you batch convert dozens of files at once while you relax. No ads, no accounts, just a fast, free tool that respects your time.

📰 https://news.ycombinator.com/item?id=50294195

🆔 @hackernewsgithubprojects

📦 dennysora/dgxtop

dgxtop: The GPU Monitor That Remembers Everything

Dgxtop keeps a permanent diary of every job your graphics card runs, so you can see exactly who ate your memory after the fact. While standard tools vanish when you close the window, this Rust-based terminal monitor logs every process start and end, including why it died. It displays your entire machine, from disk to network, but puts the GPU front and center. You see exactly which specific job holds how much memory, with colors that stay consistent across screens. It is private, lightweight, and works even if the GPU driver fails.

📰 https://news.ycombinator.com/item?id=48878070

🆔 @hackernewsgithubprojects

📦 google/three-live

Three-Live: Turn AI Code into Native 3D

Take AI-written JavaScript and turn it into native 3D apps without rebuilding your pipeline. Google released Three-Live, a runtime that compiles live Three.js code directly on your phone or headset. Instead of sticking to web browsers, it translates that code into native engines like Filament on Android and RealityKit on iOS. This means you get full augmented reality support and smooth performance that a standard browser simply cannot provide. It is essentially a bridge that lets AI agents build interactive 3D scenes that run natively on your hardware. If you want to build spatial apps without writing native code from scratch, this is your new shortcut.

📰 https://news.ycombinator.com/item?id=50352948

🆔 @hackernewsgithubprojects

📦 gauge-sh/gauge-cli

Gauge: Test Your Docs for AI Agents

Gauge runs real coding agents against your product to see where they get stuck. It captures exactly where an agent gets confused or blocked, so you can fix your documentation or tools before you publish them. The best part? You can do this entirely on your own computer. No extra accounts, no cloud costs, just your existing API keys. You change a line of help text, run the test again, and see if the agent finally succeeds. It turns guesswork into a clear, measurable loop. If you build tools that people use with AI, this is how you make sure they actually work.

📰 https://news.ycombinator.com/item?id=50371090

🆔 @hackernewsgithubprojects

📦 caldoai/skills

Skills: The AI Codebase Untangler

The single most surprising thing about the Skills repository is that it treats your entire codebase as a puzzle to be solved in one pass, rather than a pile of files to be edited. This project gives AI agents a specific method to untangle messy code, meaning it finds hidden links where pieces of software are secretly tied together and breaks them so changes stay simple. The result is that agents write better code faster and make fewer silent mistakes. It turns a chaotic project into a clean system where everything fits together naturally.

🆔 @hackernewsgithubprojects

📦 zeekk0/openamika

The Anime Girl AI Living on Your PC

OpenAmika is the local AI companion that finally lets you talk to a living character right on your desktop. It runs a realistic anime avatar that listens, speaks, and actually remembers your name and where you live. The coolest part is that she does not just chat; she can call on helper agents to search the web or control smart plugs in your home. You can run the whole brain locally on your own hardware or connect to a cloud service if you prefer. It is a genuine companion experience that feels surprisingly alive without needing a subscription.

🆔 @hackernewsgithubprojects

📦 orailnoor/cross-platform-llm-client

PrivateLM: AI That Runs On Your Phone

Your smartphone can run a large language model entirely offline. PrivateLM is a cross-platform app that executes these complex AI models directly on your device's hardware, meaning no internet connection is needed once the model is downloaded. It solves the problem of privacy and connectivity by keeping your data on your phone while still allowing you to switch to cloud providers when you need more power. This makes it interesting for anyone who wants intelligent assistance without sending personal information to external servers. The result is a private, self-contained AI experience that works wherever you are.

📰 https://news.ycombinator.com/item?id=50388641

🆔 @hackernewsgithubprojects

📦 sgaofen/humanizer-local-model

humanizer-local-model: The Local AI Rewriter That Keeps Facts

humanizer-local-model runs a twelve-billion parameter language model entirely on your own computer to rewrite AI-generated drafts so they sound like a person wrote them. The most surprising part is that it passes strict external AI detection tests ninety-five percent of the time, while keeping every single number, date, and name exactly as you wrote it. It does not just swap words for better ones; it is trained specifically to preserve factual accuracy, which means you get natural, human-sounding text without the risky hallucinations that usually come with AI editing.

🆔 @hackernewsgithubprojects

📦 calebevans/mulder

Mulder: The AI Detective That Can't Lie

Mulder is the forensic tool that uses artificial intelligence to investigate cyberattacks without ever making things up. It takes messy digital evidence like disk images and network logs, then runs a five step process to find exactly what happened. The most surprising part is a dedicated phase where an AI argues against its own findings to catch mistakes. This project matters because it solves the biggest problem with AI in security: hallucination. It forces the system to cite real evidence for every claim, making false reports impossible. The takeaway is clear.

📰 https://news.ycombinator.com/item?id=47848253

🆔 @hackernewsgithubprojects