📦 bridge-mind/bridgeclip

BridgeClip: The Open-Source Clipping App

BridgeClip turns massive video files into captioned shorts directly on your computer. It finds the best moments using AI, cuts them into the perfect aspect ratio, and burns in stylish word-by-word text. The coolest part is that everything happens locally. Your data, your keys, your machine. No middleman servers. Just a clean, open-source tool that respects your privacy while doing the heavy lifting for you. It is the smart, private way to handle your content workflow. If you create long-form video, this is the missing piece in your desktop setup. Stop letting cloud services hold your footage hostage. Get control back with BridgeClip.

🆔 @hackernewsgithubprojects

📦 edison-land/paragravity

ParaGravity

ParaGravity is the command-line tool that lets you run multiple Google Antigravity accounts simultaneously without touching the application itself. If you use AI coding agents, you know the frustration of hitting usage limits or needing separate environments for different clients. This tool solves that by creating isolated, independent copies of the app for each account. It does this safely using standard browser features, so your original installation remains untouched. Each profile gets its own folder, its own login, and its own data, meaning you can switch between personal and work projects instantly.

🆔 @hackernewsgithubprojects

📦 codejunkie99/keel

How Keel Actually Codes

Keel runs a local AI model directly on your Mac to decide how your coding agents should act. It is a native macOS workspace that unifies your editor, terminal, and multiple AI providers into one window. The interesting part is its decision layer. Instead of blindly following one model, Keel uses a local selector to pick the best route for a new task, and then a host system validates that choice before it runs. This means you get a local, private way to steer AI coding tasks without sending your code to the cloud for every decision.

🆔 @hackernewsgithubprojects

📦 fbddcz/embodied-jev

Embodied Jev: The Robot Workbench

Embodied Jev is the robot decision workbench that shows you exactly how a brain talks to a body. It lets researchers watch a digital robot arm make choices in real-time physics, comparing different thinking models side by side. You can see the arm move, check if it actually succeeds, and compare costs instantly. No guessing, no black boxes, just clear evidence of what worked and what failed. This tool turns complex robot testing into a transparent, visual experience for anyone who wants to understand how AI actually controls machines. It is the cleanest way to see robot intelligence in action.

🆔 @hackernewsgithubprojects

📦 ferrycorleone/crush-monitor

Crush Monitor

Decode how your crush really feels about you by feeding your chat history into this clever tool. Instead of overanalyzing every single line yourself, you paste your text exports from WeChat, WhatsApp, or iMessage into a clean, locally hosted dashboard. It uses a specialized structured judgment model called Jev with your own API credentials to break down emotions, categorize intentions, and even grade your own replies from SSS down to D. It scores their engagement, care, and openness over time while keeping all your sensitive conversations safely stored in your own browser's local database.

🆔 @hackernewsgithubprojects

📦 glaz-boga-poisk-bot/glaz-boga

Глаз Бога: поиск по ID и нику

Find who someone is by just their phone number, email, or Telegram ID. That is exactly what the project called Глаз Бога does. It digs through open, archived, and previously leaked records to connect those identifiers to a specific person. You type in a handle or contact detail, and it searches through massive amounts of public and old data to find matches. It is not about spying, but about seeing what information is actually floating around out there. It helps you understand how easy it is to link a digital identity to a real human being. The more data is available, the more accurate the results become.

🆔 @hackernewsgithubprojects

📦 angusdevgo/seep-reverse-lab

Seep Reverse Lab

Seep Reverse Lab runs an AI agent that audits software clients to find where they incorrectly trust the user to decide who is allowed to do what. Most security teams check the server for these authorization flaws, but this tool targets the client side, which is often overlooked and just as dangerous. It works by analyzing the compiled code of apps and binaries to spot places where a simple tweak could let a regular user act like an admin. This matters because fixing these gaps before release prevents attackers from hijacking sessions or accessing private data without proper credentials.

🆔 @hackernewsgithubprojects

📦 gufo-org/gufo

Gufo: Local AI on Strix Halo

Run massive artificial intelligence models directly on your specific laptop hardware without sending data to the cloud. Gufo is a specialized inference engine built exclusively for AMD Strix Halo processors, allowing you to process large language models, speech, and images locally. It stands out by achieving incredible speed through custom code that works directly with the chip’s unified memory, bypassing typical software layers. This means you get high performance for tasks like text generation and audio synthesis entirely on your own device. It turns a portable computer into a powerful local AI workstation. If you have this specific hardware, this tool gives you full control and speed.

🆔 @hackernewsgithubprojects

📦 haleclipse/cometixcode

Cometix Code

Rebuild a major AI coding interface in a single, fast language. Cometix Code is an unofficial Rust reimplementation of Anthropic's Claude Code terminal experience. Instead of relying on a heavier runtime, this project uses Rust to run the entire chat and code editing environment directly in your terminal. It solves the need for a lightweight, high-performance local interface that mirrors the official tool's features. The standout aspect is that it brings a complex, interactive AI workflow to the command line with the speed and efficiency of compiled code. This makes it a fascinating look at how core developer tools can be optimized from the ground up.

🆔 @hackernewsgithubprojects

📦 heman10x-ngu/openjev-verdict-2.0

The AI That Decides in 20 Milliseconds

openjev-verdict-2.0 is the lightweight decision engine that finally gets rid of the slow guessing. Instead of having a large language model spit out unstructured text that your code then has to parse and validate, this tool looks at a set of options and gives you a single, confident answer in about twenty milliseconds. It actually outperforms much larger, more expensive models while running entirely inside your web browser. That means your sensitive data never leaves your device, and you get a result you can trust. If you are tired of waiting on big models just to make a simple choice, this is the faster, smarter way to do it.

🆔 @hackernewsgithubprojects

📦 devin-axis/jev-dsh-decision

Jev DSH Decision Engine

Let your AI agent make smarter choices by plugging in a dedicated decision engine that handles the tough calls for you. This tool connects to TypeSafe AI to help your agent pick the right tools or evaluate the quality of its own work, returning clear probabilities instead of vague guesses. It works as a native plugin for DeepSeek Harness or through iPolloWork for other platforms, keeping your main agent in control while adding a layer of structured judgment. It is a neat way to add precision to your automation without rewriting your core logic. If you run AI agents, this is a solid addition to consider.

🆔 @hackernewsgithubprojects

📦 emo-gml/awesome-mental-health-llms

The Map of AI Therapy

Awesome Mental Health LLMs is the curated map that turns scattered research into one clear guide to how AI is learning to support us emotionally. It pulls together hundreds of papers showing how these models move beyond simple pattern matching to become personalized companions that can hold a conversation, remember your history, and actually help you process feelings. Instead of drowning in individual studies, you get a single, organized view of where the technology stands today. It is less about coding and more about understanding the bridge between clinical care and digital empathy.

🆔 @hackernewsgithubprojects

📦 heyjunpenn/awesome-jev

awesome-jev

awesome-jev is the community catalog that helps you discover hundreds of open-source projects built on Jev. Instead of hunting through forums, you get a verified, multilingual directory of over nine hundred tools, integrations, and models. You can easily find everything from browser automation bridges to agent tools and specialized local decision models. The site is built with Astro and runs entirely on a Cloudflare Worker, keeping the whole directory fast and easy to navigate. If you want to see what is actually possible with Jev, check out awesome-jev to find your next favorite open-source project.

🆔 @hackernewsgithubprojects

📦 huangbai-ai/post-production-skill

Post Production Skill

You can now turn simple raw footage or text prompts into cinematic retro-themed visual effects videos with flawless spatial continuity. The post-production-skill repository is a specialized toolkit built for Seedance 2.5 that creates video prompts featuring a creator in a cozy nineties studio. Instead of generating flat overlays, it maps transparent color-grading curves, CRT monitors, and adjustment panels directly into three-dimensional space with realistic depth, occlusion, and parallax. It is a brilliant way to build highly consistent, multi-shot video sequences where characters, clothing, and background objects stay perfectly locked across every single frame. You get cinematic camera work without the typical AI warping.

🆔 @hackernewsgithubprojects

📦 inkloom-art/inkloom

Inkloom

Most artificial intelligence generators produce logos that are just flat, unconstructed pictures with messy letters and zero geometry. Inkloom is changing that by building specialized models that design a logo the exact way a professional studio does. Instead of guessing, it analyzes your brand to set strict geometric constraints, optically spaces and aligns the custom typography, and builds symbols using precise mathematical rules. The output is a real specification rather than a random bitmap, giving you a mark you can actually describe, defend, and send to a printer. Check out their open-source repository to see how they are building this incredibly clean platform.

🆔 @hackernewsgithubprojects

📦 hynetworks/opengfw

OpenGFW

OpenGFW is the open source Linux tool that lets you run your own personal version of the Great Firewall of China right on your home router. Originally built for research, ad blocking, and parental controls, this powerful system analyzes network traffic in real time. It can identify and filter complex protocols like secure web traffic, encrypted VPNs, and even hidden proxy tunnels that standard blockers usually miss. While governments use these methods for national censorship, OpenGFW puts that same sophisticated technology in your hands to secure your home network, block tracking, and explore how large-scale filtering actually works behind the scenes.

🆔 @hackernewsgithubprojects

📦 jameskinner/covalent-mas

Covalent MAS

Covalent drug design just got a massive upgrade with a new open-source pipeline that boosts valid molecular generation rates up to nearly ninety-nine percent. It is called covalent-mas, and it provides a clever, modular framework that connects every stage of covalent drug discovery into a single learning loop. Instead of manually hacking together generative models, docking tools like AutoDock Vina, and custom filters, this tool standardizes everything through neat Python interfaces. The coolest part is how it records every single run as a trajectory event, automatically building traceable memories and skills to guide your next round of molecular optimization.

🆔 @hackernewsgithubprojects

📦 jev-chat/jev-chat-jarvis

jev-chat-jarvis

jev-chat-jarvis is the Android companion app that finally acts as a personal copilot for your daily chat applications. Unlike intrusive tools that modify or hack your messaging apps, this clever assistant runs safely in the background using Android accessibility features to read your screen. It analyzes the incoming message to figure out the sender's true intent and risk level, drafts three candidate responses, and lets you paste your choice with a single tap. It never sends anything automatically, keeping you in complete control. Give it a spin to take the stress out of busy messaging days.

🆔 @hackernewsgithubprojects

📦 jev-chat/jev-chat-jarvis-mac

Jev Chat Jarvis for Mac

A tiny app on your Mac reads the chat window in front of you without touching the database or taking screenshots. It uses a local model to guess what a message really means and how risky it is, then drafts a few replies you can paste into the chat. This is Jev Chat Jarvis for Mac, a floating helper that stays out of your way while you decide what to send. It is strictly read-only, so your privacy stays yours. The real surprise is that it runs entirely on your computer, so your conversations never leave your machine unless you choose to send them.

🆔 @hackernewsgithubprojects

📦 joeseesun/qiaomu-download

The Video Downloader That Actually Checks Your Files

Qiaomu Download is the video downloader that refuses to call a job done until the file actually plays. Most tools just grab a file and hope for the best, but this one runs a deep check to make sure your video isn't corrupted before you even know it arrived. It handles links from YouTube, TikTok, and even Spotify tracks by finding the matching official audio, all while keeping your private login data locked down. If you are tired of guessing whether a download is good, this gives you a clear green light so you know your media is ready to use.

🆔 @hackernewsgithubprojects

📦 kryvora-network/kryvora-node

Kryvora Node

Your computer is now a paid participant in a decentralized physical network. Kryvora Node runs a background daemon that connects your hardware to a central coordination hub, allowing it to run verification tasks and prove its availability to earn rewards. It is the official client for a DePIN network, handling everything from telemetry probes to peer state checks. To get started, you need a specific on-chain credential called a Genesis Node Key to authenticate your worker. Once connected, your machine becomes part of a distributed system that verifies real-world infrastructure, turning idle compute resources into active, verified nodes.

🆔 @hackernewsgithubprojects