📦 kartiklabhshetwar/better-shot

BetterShot

BetterShot is a completely open-source, local-first alternative to CleanShot X built natively for macOS with Swift and SwiftUI. It replaces the default Mac screenshot utility with a fast, lightweight tool that runs entirely on your machine without subscriptions, cloud storage, or telemetry tracking. When you capture a window or record your screen, the app lets you instantly polish the visual with custom background gradients, adjustable padding, and clean corner shadows. It also includes built-in tools for on-screen color picking, text extraction, and quick canvas annotations. It is the perfect native utility for developers who want beautiful, professional screenshots without sacrificing their privacy.

🆔 @hackernewsgithubprojects

📦 gowrav-vishwakarma/qllm2

qllm2

qllm2 is the attention-free language model that keeps your context footprint completely flat while retaining massive associative memory. Instead of relying on heavy transformers or standard state-space models, this project introduces phase-associative memory, where tokens live in a complex phase space. By replacing the traditional growing key-value cache with a clever complex-conjugate matrix-state layer, it achieves constant time-per-token inference. This means you get a model that can recall deep associations without dragging around an ever-expanding memory burden. It is a brilliant rethink of how sequence modeling can work using pure math instead of massive hardware brute force.

🆔 @hackernewsgithubprojects

📦 shivfun99/pulse-order

pulse-order

The C++20 packet processor that brings ultra-low latency trading logic right down to the bare metal. By cutting out heavy databases and web APIs, pulse-order focuses entirely on the critical path of trading. It grabs raw Ethernet frames, decodes them instantly, updates an in-memory order book, runs a quick strategy check, and sends order packets back out using DPDK. By avoiding dynamic memory allocation and using fixed-size binary data, it handles this entire sequence in just over one hundred nanoseconds on a virtual link. It is a brilliant masterclass in writing incredibly fast, predictable code by simply avoiding unnecessary work.

🆔 @hackernewsgithubprojects

📦 agentseal/codeburn

codeburn

codeburn is the local-first CLI tool that finally tracks your AI coding token usage and costs across thirty-one different tools. Instead of staring at a mysterious, massive monthly bill, you just run a single command to see exactly where your budget is vanishing. It parses local session files from tools like Claude Code and Cursor right on your machine, breaking down every single penny by task, model, and project. You can instantly spot wasted tokens on redundant context or compare models side-by-side to see which one actually gets your code shipped. Grab codeburn on npm to finally take control of your AI development budget today.

📰 https://news.ycombinator.com/item?id=47759035

🆔 @hackernewsgithubprojects

📦 researai/autofigure-edit

AutoFigure-Edit

Autofigure-edit is the open-source pipeline that finally turns raw academic paper text into fully editable vector graphics. Instead of forcing you to settle for rigid static images, this tool uses AI to draft a schematic, runs smart segmentation to isolate individual icons, and automatically builds a beautifully structured SVG file. You can easily tweak shapes, swap colors, and rewrite labels directly in an embedded browser editor, or even feed it a reference image to copy a specific journal style. It is a brilliant way to build professional, custom science diagrams without starting from scratch.

🆔 @hackernewsgithubprojects

📦 nicolasvonluetzow/gaussiangpt

GaussianGPT

While most 3D generators rely on complex diffusion pipelines, gaussiangpt creates entire 3D scenes step-by-step just like a text-writing chatbot. By compressing 3D Gaussian primitives into discrete tokens, it uses a transformer to predict the next spatial piece of a scene sequentially. This clever autoregressive approach solves the rigid limitations of holistic generation, making it incredibly easy to complete half-finished rooms or expand existing environments through simple outpainting. It is a highly practical and flexible tool for anyone looking to build expansive virtual worlds. Check out the repository to start generating, completing, and expanding your own detailed 3D scenes today.

🆔 @hackernewsgithubprojects

📦 toolsai/notebooklm-studio-skill

notebooklm-studio-skill

You can now connect Google NotebookLM to your local AI agent for a fully automated research workflow. The notebooklm-studio-skill repository turns your AI assistant into an automated research operator that handles the entire process end to end. Instead of manually clicking through the web interface, this tool lets your agent batch-upload diverse sources, trigger complex generation jobs, and automatically download the results to your local machine. It even converts raw mind map data into interactive web pages and launches a local dashboard to track your research jobs. It is the perfect bridge for turning raw documents into structured insights.

🆔 @hackernewsgithubprojects

📦 tahash/wayflow

Wayflow

Drop a complete visual workflow editor directly into your web app with just a single function call. You get an interactive canvas, a node palette, a configuration panel, and run controls immediately without building anything from scratch. Because it is written in plain TypeScript and standard web technology, it integrates perfectly into any framework, from React and Vue to no framework at all. It even includes a built-in execution engine to run your logic in the browser or on your server. You can mix traditional step-by-step logic with smart AI actions, and pause workflows to wait for human approval before resuming.

🆔 @hackernewsgithubprojects

📦 lbouaraba/comfyui-krea2edit

comfyui-krea2edit

You can now edit specific parts of an image while perfectly preserving the identity of your subject. A new custom node pack called comfyui-krea2edit brings the Krea 2 Identity Edit model directly into your ComfyUI workflow. Instead of guessing, the system uses dual conditioning, feeding your original image to both the latent decoder and the text encoder. This means the AI actually sees the image semantically while reading your editing prompt. It is a game-changer for making complex edits, like swapping out a background or changing clothes, without altering the face or structure of your original subject.

🆔 @hackernewsgithubprojects

📦 roversx/launchnext

LaunchNext

Bring back your classic macOS app launcher and escape the limitations of newer system updates. This open-source tool instantly recreates your familiar Launchpad layout, including all your custom folders and app positions, by reading directly from your system database. You get full control to drag apps around, create custom folders, and even use keyboard fuzzy search to open programs in a flash. It even supports hot corners and trackpad gestures to launch your apps exactly how you like. Grab this free utility to reclaim your desktop workspace and keep your Mac running exactly the way you want.

🆔 @hackernewsgithubprojects

📦 nanako0129/pilotfish

pilotfish

Anthropic's latest tests show a smart orchestrator paired with cheap worker models delivers ninety-six percent of full-frontier performance for under half the cost. That is exactly what pilotfish automates for Claude Code. This clever zero-runtime configuration layers on top of your global settings, putting a premium model in your main session to handle the high-level planning while routing routine tasks like searching, mechanical edits, and test runs to cheaper subagents like Sonnet and Haiku. It even gates non-trivial work behind an independent verifier agent to catch errors before they hit your code. It is the ultimate way to stretch your weekly subscription limits.

🆔 @hackernewsgithubprojects

📦 nytemodeonly/polyterm

polyterm

Track prediction markets directly from your command line without dealing with heavy web browsers or complex interfaces. This terminal application serves as a dedicated intelligence layer for Polymarket, turning raw data into actionable insights. Its standout feature is an arbitrage scanner that automatically matches prices between Polymarket and Kalshi, pointing out immediate discrepancies and guaranteed profit opportunities. You can easily analyze order books, calculate potential returns with built-in calculators, and monitor big-money whale wallets, all inside a beautifully structured interactive terminal screen. It is a completely view-only tool, keeping your personal keys safe while you research.

🆔 @hackernewsgithubprojects

📦 bysiber/cleardisk

ClearDisk

Your Mac is likely hiding up to five hundred gigabytes of forgotten developer caches, but a tiny tool called cleardisk can find and clear them in seconds. This lightweight, open-source macOS menu bar app monitors sixty-three different developer cache paths—covering everything from Xcode and npm to Docker and Cargo. It even scans your project folders for stale modules and build files, letting you safely reclaim massive amounts of SSD space by sending them straight to your trash. It is completely free, runs with zero dependencies, and never requests network access, giving you a private, clever way to keep your machine running light.

🆔 @hackernewsgithubprojects

📦 pwilkin/openmoss

OpenMOSS

Run voice cloning and high-quality speech synthesis entirely on your own local hardware using a single self-contained tool. With openmoss, you get a clean C++ translation of a massive audio model pipeline that runs directly on your computer without any complex external dependencies. This project packs a powerful language backbone and an advanced audio codec together, letting you generate realistic speech or copy a target voice using just a short reference audio file. It is incredibly efficient, producing ten seconds of natural speech in just four seconds on a standard graphics card. Grab this tool to run blazing fast local voice generation completely under your control.

🆔 @hackernewsgithubprojects

📦 wassermanproductions/blockout

Blockout

AI video generators produce dramatically better clips when you hand them an unambiguous motion reference. Blockout is a clever desktop app that lets you stage grey-box 3D scenes, choreograph camera moves, and block out character paths using a simple timeline and physical marks. It is deliberately not a 3D art tool, focusing entirely on real-world optics, timing, and choreography. When you are done blocking, it exports a synchronized package of reference videos, depth passes, and tailored camera prompts that models like Veo, Kling, and Wan can follow precisely. It is a fantastic way to make sure your AI filmmaker workflow actually shoots the angle you wanted.

🆔 @hackernewsgithubprojects

📦 rvp-group/gsolver

gsolver

The distributed mapping framework that seamlessly blends continuous-time trajectory estimation with decentralized message passing, gsolver is a powerful tool for robotics developers tackling complex spatial mapping tasks. Traditional robot mapping struggles when multiple machines need to coordinate or when camera shutters distort moving images. This project solves that by combining continuous-time motion models with localized belief propagation, letting robots update their maps cooperatively without a single centralized computer. The coolest part is that it offers fully continuous trajectories, meaning you can query the robot's exact position at any fractional millisecond. Check out gsolver to bring decentralized, real-time spatial mapping to your next robotics setup.

🆔 @hackernewsgithubprojects

📦 zhou-weichen/geosubprobe

Geosubprobe

Geosubprobe is the diagnostic toolbox that finally reveals exactly how self-supervised vision models understand the physical, three-dimensional geometry of the images they see. Instead of treating these advanced vision models like black boxes, this tool trains specialized probes to dissect their internal layers and analyze how they represent spatial concepts like depth and surface angles across different datasets. By inserting targeted interventions directly into the model's internal subspaces, researchers can pinpoint where 3D awareness actually forms during training. It is an invaluable way to audit and compare different vision backbones to see which ones build the most accurate mental map of our physical world.

🆔 @hackernewsgithubprojects

📦 yuzu-hub/appscreen

Appscreen

Appscreen is the browser-based design tool that finally makes creating gorgeous App Store screenshots incredibly simple. Instead of wrestling with complex graphic design software, you just drag and drop your raw app images directly into your browser. The coolest feature is its interactive 3D mode, which lets you wrap your screenshots around an iPhone mockup and tilt, rotate, or position it in 3D space with simple mouse drags. You can easily customize gradients, add polished text overlays, and even translate your marketing copy into multiple languages automatically using built-in AI. It is the perfect, hassle-free way to build professional app store graphics in minutes.

🆔 @hackernewsgithubprojects

📦 mims-harvard/athena

Athena AI

Athena is the medical AI assistant that actually researches patient treatment decisions before answering. Instead of relying on raw memorization or guessing like typical chatbots, this system connects directly to a digital library of over two hundred validated biomedical databases and tools. When you ask a complex medical question, the agent takes a step-by-step approach, systematically looking up drug interactions, clinical warnings, and patient dosage guidelines to build its logic. It is so effective that it outperforms massive commercial AI models on treatment-planning tests. It is a fantastic tool for researchers wanting to see how specialized AI can safely reason through complex science.

🆔 @hackernewsgithubprojects

📦 rohitg00/tailclaude

TailClaude

TailClaude is the secure bridge that finally frees Claude Code from the terminal and brings it straight to your web browser. Instead of squinting at SSH terminals on a tiny phone screen, this clever tool lets you publish your coding assistant interface to any device over a secure connection. You can instantly access a beautiful, touch-friendly UI complete with streaming markdown answers, model switching, and real-time cost tracking just by scanning a QR code on your terminal. It is the ultimate way to code on the go without sacrificing the visual comfort of a modern web application.

🆔 @hackernewsgithubprojects