📦 k-dense-ai/agentic-data-scientist

Agentic Data Scientist

Agentic Data Scientist lets you automate complex data analysis by using multiple AI agents to plan, code, and review your work end to end. Instead of just writing a script, it breaks your request into clear steps, writes the code, and then checks its own work against strict success criteria. If a step fails, it fixes the plan and tries again, ensuring the final result is actually correct. The most interesting part is how it adapts during execution, changing its strategy when it discovers new insights in your data.

🆔 @hackernewsgithubprojects

📦 will-ness-ai/skills

Skills: Real Engineer Agent Tools

The skills repository lets you install reusable agent behaviors directly into your coding environment. It turns complex workflows into simple commands that an AI assistant can execute. The most interesting feature is the flashlight tool, which helps you explore new ideas without losing track of your main project map. Instead of following a rigid plan, you can shine a light into unknown territory, gather facts, and update your project chart based on what you learn. This means you can pivot quickly when you discover better approaches, keeping your work flexible and focused. It is a practical way to keep your development process organized while staying open to new directions.

🆔 @hackernewsgithubprojects

📦 owngoalstudio/trollvnc

Control Your iPhone Screen From Anywhere

Take remote control of your iOS device directly from a web browser. TrollVNC turns your iPhone into a VNC server, so you can see the screen and tap buttons from your computer without installing special software. It even lets you copy and paste text between your phone and computer. The best part is the built-in web viewer, which means you can connect from Chrome or Safari on any device. Just point your browser at the address, and you are instantly managing your phone. It is the easiest way to get remote access when you need it.

🆔 @hackernewsgithubprojects

📦 scalingintelligence/trace

Trace: The AI Agent Trainer That Fixes Specific Weaknesses

Trace is the AI agent trainer that automatically fixes specific weaknesses instead of just throwing more data at a problem. It works by watching an AI fail, spotting exactly which skill it is missing, and then creating custom practice exercises just for that one gap. The system builds tiny training worlds to isolate the issue, teaches the model using specialized lightweight updates, and finally adds a smart router that instantly picks the right fix for whatever task comes next. This means you get a sharper, more reliable agent without wasting time on things it already knows. It is a clever way to make AI agents actually learn from their mistakes.

🆔 @hackernewsgithubprojects

📦 hamelsmu/research-council

Research Council: AI Agents That Debate Each Other

Run the research council plugin and watch two coding assistants debate your topic simultaneously. One assistant from Claude explores the angles while a second from OpenAI does the same, and then they actually swap notes. They read each other’s drafts, challenge weak points, and dig deeper to fix blind spots before combining everything into one solid report. This cross-pollination step is the magic because different AI models think differently, so you get a richer, more balanced answer than any single bot could produce alone.

🆔 @hackernewsgithubprojects

📦 themuffinator/openq4

Play Quake 4 on Modern PCs with openQ4

Run the classic Quake 4 shooter on today's computers with openQ4, a free open-source engine that breathes new life into this beloved title while keeping all the original story and levels intact. You simply drop this project into your existing game folder to replace the outdated binaries with a modern engine that finally supports widescreen displays, ultrawide monitors, and high-resolution textures without breaking the original experience. The developers have carefully preserved the authentic retro feel by using the retail assets you already own, but they added smart upgrades like modern lighting, bloom effects, and smoother controls that make the game feel crisp on current hardware.

🆔 @hackernewsgithubprojects

📦 hitsz-tmg/lychee-fd

Lychee-FD: The Full-Duplex Spoken Language Model

Hear a conversation that flows like real life without awkward pauses or waiting for the computer to finish speaking. Lychee-FD is a new spoken language model that listens and talks at the same time just like two people chatting together. Traditional voice assistants force you to wait for a beep before you can reply which feels stiff and unnatural. This project breaks that mold by handling continuous listening and speaking within one unified system. It understands your words while simultaneously generating a natural vocal response and even knows when to interrupt you to keep the dialogue moving forward.

🆔 @hackernewsgithubprojects

📦 antoniorodr/cronboard

Cronboard

Cronboard lets you manage scheduled tasks from a friendly terminal dashboard. Instead of editing cryptic config files by hand, you get a clean interface to add, pause, or delete jobs on your computer or remote servers. The coolest part is the real-time logging system. You can watch exactly what each scheduled task does as it runs, spotting errors immediately instead of wondering why a job failed hours later. You can even set it to ping your phone via Telegram if something breaks. It makes server maintenance way less stressful and actually fun.

🆔 @hackernewsgithubprojects

📦 lovdacn/ui

LovdaCN: Shadcn for React Native

LovdaCN is the component library that brings the popular Shadcn philosophy of copy-paste customization directly to React Native apps. Instead of locking you into a rigid black-box library, it installs clean, full-source components into your project, letting you tweak every detail without fighting framework constraints. What makes it genuinely useful is that it solves the headache of styling in mobile development by supporting modern CSS systems like NativeWind and Uniwind out of the box. You simply run a command to grab buttons, inputs, or dialogs, and they land in your codebase ready to be molded to your brand.

🆔 @hackernewsgithubprojects

📦 ryanbr/noop

Noop

Noop is the offline companion app that finally lets you control your own biometric data without paying for a subscription. By pairing directly with your fitness strap over Bluetooth, it processes all of your sleep, recovery, and strain metrics locally on your computer or phone. Instead of sending your sensitive health information to corporate servers, everything is kept in a private database on your own device. You can even import your historical data, set up custom haptic alarms, or use the physical double tap on your band to trigger actions on your computer. It is a completely private, account free way to own your health stats.

📰 https://news.ycombinator.com/item?id=49280764

🆔 @hackernewsgithubprojects

📦 memorax-ai/memorax-code

MemoraX Code: AI Memory for Developers

MemoraX Code lets AI coding agents remember your past engineering decisions, repository structure, and personal preferences across separate sessions. It stops agents from starting over by recalling verified fixes, failed attempts, and team workflows automatically. You keep control by viewing and editing memories in a local dashboard or cloud console, while the plugin works in the background to reduce repetitive search and trial-and-error. It is a practical way to make your coding sessions compound in value instead of resetting every time you open a new chat.

🆔 @hackernewsgithubprojects

📦 tobi/omasnap

Omasnap: Screenshot Tool with Pinned Overlays

Omasnap lets you pin a screenshot permanently to your desktop as a floating layer, something most screenshot tools simply ignore. Built for Wayland users on Hyprland or Omarchy, it captures screens with clean native resolution and lets you annotate them using fast vector layers like arrows, text, and highlights. What makes it unique is the pin feature, which anchors a finished capture to the bottom-right corner of your screen, keeping it visible across all workspaces without requiring extra system rules. You can resize, edit, or drag that pinned image directly into other apps while it stays right there.

🆔 @hackernewsgithubprojects

📦 dmmulroy/anti-slop

Anti-Slop: Stop Lying to Your TypeScript Types

Stop lying to your TypeScript types with anti-slop. This clever Oxlint plugin aggressively bans weak patterns like chained type assertions and unsafe dictionary types that hide real errors. Instead of letting vague code slide, it forces you to prove what your variables actually hold at every step. You get tighter, safer code without the usual guesswork. It is the perfect tool for developers who want their types to mean something real. Give your code some actual substance.

📰 https://news.ycombinator.com/item?id=49272634

🆔 @hackernewsgithubprojects

📦 uditakhourii/neuroarxiv

NeuroArxiv: Stop Guessing New Tech

NeuroArxiv is the coding skill that forces AI to check real research papers before inventing new solutions. Most tools just throw general search results at a problem, but this project makes Claude fetch specific arXiv papers first. It reads each abstract in isolation so the model doesn’t get distracted by other sources. Then it picks exactly one path to take, cites the winner, and warns you about known mistakes from similar work. This stops you from building something that was already tried and failed. It keeps your architecture grounded in actual science instead of generic advice.

🆔 @hackernewsgithubprojects

📦 mobai-app/ios-builder

Build iOS Apps Without a Mac

Build iOS apps from Windows or Linux without needing a Mac. Builder uses GitHub Actions to compile your code on Apple servers and sends the finished app straight to your computer. The coolest part is testing it instantly: you can tap through your new app in a simulator right from your Windows desktop, or even hot reload changes on a real iPhone while sitting on Linux. It basically bridges the gap so you can develop the whole ecosystem without touching Apple hardware.

🆔 @hackernewsgithubprojects

📦 jerryzliu/dayflow

Dayflow: The Private Mac App That Writes Your Work Journal

Build a private work journal on your Mac without lifting a finger or setting timers. Dayflow quietly watches your screen activity and turns those moments into a clear, chronological timeline of what you actually accomplished. Instead of just telling you which app was open, it understands the context, so it knows whether you were coding, debugging, or just browsing. You get a daily summary ready for standups, a weekly review of where your time went, and even a chat feature to ask questions about your past work. Everything stays local on your machine, so your screen history never leaves your computer.

🆔 @hackernewsgithubprojects

📦 lcharvol/macsift

macsift: The Honest Mac Cleaner

Stop trusting mystery cleaners that nuke your system blindly. This project lets you see exactly what is eating your hard drive by grouping thousands of scattered files into neat, readable buckets based on which app created them. Instead of deleting things instantly, it moves everything to the Trash so you can click around, double check, and change your mind. It even spots old iPhone backups and developer cache files that usually hide in plain sight. You get full control over every gigabyte without fear of permanent loss. Take back your storage with total peace of mind.

🆔 @hackernewsgithubprojects

📦 alexgreensh/attention-span

Attention Span: ADHD-Friendly AI Output

Attention Span changes how Claude Code talks to you, not what it codes. It replaces walls of text with an ADHD-friendly style that puts the answer in the first line and uses bold keywords you can skim. A real benchmark shows it cuts output by forty percent while keeping coding accuracy identical, so you read faster without losing quality. It also includes a terse Spartan mode for pure signal and a briefing style with checkboxes for status updates. You install it by dropping a single markdown file into your styles folder and switching it on, then your AI assistant finally respects your attention instead of burning tokens.

🆔 @hackernewsgithubprojects

📦 shift-labs-ai/pi-peer

Pi-Peer: Let AI Sessions Talk to Each Other

Pi-peer is the messaging layer that lets your AI coding sessions find and chat with each other. Instead of copy-pasting notes between terminals, you simply tell one session to message another, and it delivers plain text right into the other’s workflow. It works locally on your machine and scales across your team using a shared server, so you can alert a colleague that a branch moved or ask if anyone is mid-task. The design is intentionally simple: it sends only text, keeps a persistent mailbox for offline messages, and strictly prevents one session from controlling another. It turns isolated AI agents into a cooperative team without adding complexity.

🆔 @hackernewsgithubprojects

📦 mateusdcc/pi-gpt-search

Pi-Gpt-Search: Zero Cost Web Search for Pi

Pi Gpt Search lets any coding model grab fresh web results without burning a single token of your model’s quota. Instead of making the AI guess or wait, it quietly taps into OpenAI’s own search backend to pull live snippets and links directly. Your main model stays in charge of thinking, while this tool handles the fetching in the background. It even supports deep multi-step research that follows links and checks page content automatically. You get current facts without paying extra for inference or reading stale data. It works with any model you choose and keeps your privacy intact by only sending the search query.

📰 https://news.ycombinator.com/item?id=49255837

🆔 @hackernewsgithubprojects

📦 jiashunwang/hybrid-motion-imitation

hybrid-motion-imitation

hybrid-motion-imitation lets robots handle both walking and climbing by mixing two distinct movement styles into one smooth system. Instead of learning each action separately it trains a single agent that switches between locomotion and whole-body tasks based on terrain. You can use real motion capture data to teach the robot how to climb boxes or grab objects then watch it adapt in simulation or on actual hardware. It builds on proven research but makes it easier to run with standard simulators. This approach means your robot does not just shuffle around but actually interacts with the world around it making complex tasks feel natural.

🆔 @hackernewsgithubprojects