📦 russellenvy/three-man-team

Three Man Team

Prevent your AI coding assistants from burning through tokens and drifting mid-task by organizing them into a structured, three-agent development team inside a single session. This project sets up an Architect to plan and deploy, a Builder to write the exact code needed, and a Reviewer to verify the quality before anything ships. Instead of wasting resources on massive context windows, these specialized personas hand off tasks using simple markdown files. It keeps the AI disciplined, focused on one change at a time, and running efficiently inside your terminal. You get a reliable, repeatable workflow that turns erratic AI prompts into a structured software shipping pipeline.

🆔 @hackernewsgithubprojects

📦 nvidia/audio2face-3d

audio2face-3d

audio2face-3d is the open collection that generates realistic 3D facial animation straight from any audio file or live stream. Instead of spending hours manually animating mouth movements, this project acts as a central hub of tools, models, and plugins to turn voice data into natural motion. It analyzes the sound to automatically calculate precise jaw movement, subtle skin deformations, and even emotional expressions. You can run it directly in game engines like Unreal Engine 5 or design tools like Autodesk Maya. It is a fantastic way to bring digital characters to life instantly using real-world speech.

🆔 @hackernewsgithubprojects

📦 videlalvaro/leet-llm

Leet LLM

Build your own custom AI engine directly on your Mac by coding it from scratch. Leet-llm is a hands-on, forty-seven lesson course designed to teach you how large language models actually run on Apple Silicon. Instead of just reading theory, you write Swift code for the CPU and Metal code for the GPU, piece by piece. You start with basic math operations like vector dot products and gradually assemble them into a working token generator. It features its own native macOS app with visual diagrams and a massive companion book, making complex GPU programming feel incredibly approachable. It is the ultimate playground for understanding how AI hardware really works.

🆔 @hackernewsgithubprojects

📦 jenissimo/bottleship

Bottleship

You can now run classic 3D Windows games directly in a web browser without installing any plugins. An incredible project called bottleship is a high-level emulation engine that boots real x86 Windows executables right inside a web page. Instead of using slow, heavy virtual machines, it translates vintage system calls on the fly. It actually rebuilds original graphics, audio, and file systems using modern browser technologies like WebGPU and WebAudio. This means iconic games from the nineties and early two-thousands can load and render with hardware acceleration on almost any modern device. It is a brilliant way to keep classic PC gaming history alive and instantly playable.

🆔 @hackernewsgithubprojects

📦 juggler-ai/juggler

Juggler Code Agent

Juggler is the AI coding agent that finally gives you a visual workbench instead of a linear chat window. Most coding assistants bury tool calls and conversational threads in a single endless doom-scroll. Juggler flips the script by turning your entire coding session into an interactive tree, elegantly displayed in Finder-style columns. You can branch off into sub-threads, backtrack, and edit the context directly, while the headless Go server runs locally or remotely. It is a highly customizable, collaborative playground where you can easily inspect and edit every prompt and tool call. Try it out to take real control of your AI workflow.

📰 https://news.ycombinator.com/item?id=48883305

🆔 @hackernewsgithubprojects

📦 othmanadi/plandeck

Plandeck

Keep your long-running AI coding agents on track by visualizing their progress with a live Kanban board that lives on your local disk. Instead of letting your agent get confused after a context reset, plandeck turns flat markdown files into a structured, queryable task graph. It automatically promotes cards to ready as dependencies are met, lights up the critical path in gold, and outputs a tiny next-action file so your agent instantly knows its next move. It is powered by a reliable, zero-dependency engine that reads plain text, giving you clear tracking without any complicated setups.

🆔 @hackernewsgithubprojects

📦 flyfish-dev/file-viewer

file-viewer

file-viewer is the browser-native document previewer that finally eliminates the need for expensive, clunky server-side file conversion. Instead of forcing you to build and run a separate rendering server for every file type, this clever tool parses and renders over two hundred file formats directly in the user's web browser. It handles everything from standard PDFs and Office documents to CAD drawings, zip archives, and 3D models. Because it runs entirely on the client side, your private files stay secure inside your local network without ever leaking to third-party services. It integrates beautifully with React, Vue, Svelte, or vanilla JavaScript. Drop it into your app and enjoy instant, secure previews.

🆔 @hackernewsgithubprojects

📦 jayantdevkar/claude-code-karma

Claude Code Karma

Claude Code silently deletes your local coding history after thirty days, burying your precious session data in obscure system files. Fortunately, claude-code-karma rescues this data and turns it into a private, local-first visual dashboard on your own machine. It instantly indexes your terminal activity to display gorgeous timelines, token costs, spawned subagents, and file edits. You can even automatically link sessions directly to your GitHub, Jira, or Linear tickets based on your git branch names. It is the ultimate way to visualize your coding rhythm and see exactly what your AI companion has been doing.

🆔 @hackernewsgithubprojects

📦 danneauxs/pocket-tts-spokenword

pocket-tts-spokenword

pocket-tts-spokenword is the desktop-friendly audiobook generator that turns dry digital voices into emotionally expressive narration. While typical text-to-speech programs sound robotic and flat, this tool analyzes your text using advanced AI emotion classification to automatically adjust vocal tones, pacing, and feeling. It intelligently divides long texts at natural sentence boundaries, maps six distinct emotional states like joy or sadness to the speech generator, and even lets you clone custom voices. Because it runs incredibly fast using only a standard computer processor, you can easily convert entire books into high-quality, expressive audio files right from a simple, interactive desktop app.

🆔 @hackernewsgithubprojects

📦 hexxt-git/ccpool

ccpool

Sharing a single Claude Code subscription with your team usually ends in disaster when someone quietly burns the entire weekly limit by noon and nobody knows who did it. ccpool solves this by giving your team a live, shared terminal dashboard that breaks down exactly who is driving the account's usage. By running a lightweight background daemon on each member's machine, it securely correlates your local coding activity with Anthropic's official usage limits. It is a read-only observer that displays beautiful, real-time usage bars and countdowns for the group, making fair use visible and easy to coordinate.

🆔 @hackernewsgithubprojects

📦 modalitydance/palmclaw

PalmClaw

You can now run a fully functional, automated AI agent directly on your Android phone without needing a computer. An open-source project called palmclaw packages an entire agent stack—including memory, tools, and automation—into a single mobile app. Instead of relying on complex desktop setups, it operates inside Android's secure app sandbox to keep your private data safe. It connects with local files, hardware like Bluetooth, and external communication channels. You get a private, safe assistant that can automate tasks and handle tools right in your pocket. Grab the app to put a capable agent on your phone.

🆔 @hackernewsgithubprojects

📦 tencent-hunyuan/hy-embodied

hy-embodied

Imagine a robot that can navigate and interact with the physical world using an incredibly efficient brain that only activates three billion parameters at a time. The hy-embodied repository introduces a vision-language foundation model designed specifically for real-world robotic agents. It solves the massive computing bottleneck of physical deployment by utilizing a clever mixture-of-experts architecture, giving robots high-level reasoning and physical awareness without needing a giant server farm. This means robots can now plan complex tasks, adapt to changing situations, and understand spatial coordinates in real time. It is a massive step toward making smart, responsive physical robots actually practical.

🆔 @hackernewsgithubprojects

📦 scottstts/threejs-awesome-graphics-agent-skills

threejs-awesome-graphics-agent-skills

Artificial intelligence models are notoriously bad at generating high-end 3D graphics because they do not understand the math behind good taste. The threejs-awesome-graphics-agent-skills repository changes this by acting as a specialized library of expert graphics techniques designed specifically for AI agents. Instead of trying to teach an LLM basic code, this tool injects production-ready math, shaders, and rendering pipelines directly into the agent's vocabulary. It allows AI coding assistants to instantly implement complex visual effects like realistic spectral oceans, volumetric clouds, and advanced lighting models without falling back on cheap visual hacks. It is a brilliant way to make AI-generated 3D scenes actually look beautiful.

🆔 @hackernewsgithubprojects

📦 smileformylove/xscientist

xscientist

This project transforms scientific research into a git-like protocol where experiments are completely diffable, forkable, and reviewable just like open-source code. Instead of simply generating a static paper, xscientist runs a continuous pipeline that treats every research attempt as a trackable commit, mapping out your entire scientific exploration as a branching tree of code, metrics, and failures. It even links written manuscript claims directly back to the exact code execution node that proved them. It is basically git for science, letting you branch off previous experiments to build a fully verifiable, automated tech tree.

🆔 @hackernewsgithubprojects

📦 ireliya/terralogic

Terralogic

Terralogic is the benchmarking framework that finally helps artificial intelligence make sense of Earth observation data by coordinating complex, multi-step search missions. Instead of just looking at a satellite image, this system acts like a coordinator for specialized remote sensing tools across optical, radar, and infrared sensors. It breaks down complex natural language requests, builds a logical execution plan, and runs detailed analysis like counting specific structures or tracking small infrared targets over time. If a sensor tool fails or returns bad data, the system automatically self-corrects and swaps in a compatible tool to finish the job, translating raw data into verified, plain-English answers.

🆔 @hackernewsgithubprojects

📦 metainfer/metainfer

MetaInfer

Build a custom, high-performance language model inference engine from scratch using an automated, agent-driven compiler approach. Instead of downloading a massive, bloated framework with endless configuration options, this project introduces a brilliant concept called Large Language Model as Compiler. It uses a structured knowledge graph and a multi-agent system of coders, reviewers, and testers to automatically assemble a lean, single-path inference engine tailored precisely to your specific model and hardware. This eliminates thousands of lines of dead code, resulting in massive speedups over traditional setups. It is a fascinating glimpse into the future of self-assembling, hyper-optimized AI infrastructure.

🆔 @hackernewsgithubprojects

📦 jetxu-llm/codex-deepseek-bridge

Codex DeepSeek Bridge

Redirect your OpenAI Codex app requests to DeepSeek using a clever local proxy that keeps your existing setup completely intact. This tool hooks directly into your configuration so you can use deepseek-pro and deepseek-flash while keeping your plugins, MCP servers, and custom workflows running exactly as they are. The best part is your API key never leaves your machine, and you get a beautiful offline dashboard to track your cache hits and latency. When you want to revert, a single restore command puts everything back to the official defaults without leaving a trace.

🆔 @hackernewsgithubprojects

📦 thudm/scale-cua

ScaleCUA

Teach your AI models to control any desktop operating system like a human. This project introduces a pipeline that generates thousands of realistic computer tasks, like browsing the web or editing spreadsheets, and builds automated judges to verify if the AI actually completed them correctly. Instead of wasting time on tasks that are too easy or too hard, it uses a smart sampling method to constantly feed the AI challenges right at the edge of its current capabilities. It is a highly efficient way to train digital agents, making them smarter and much faster at navigating real software.

🆔 @hackernewsgithubprojects

📦 starnodes2024/comfyui-starnodes-modelconverter

comfyui-starnodes-modelconverter

You can now squeeze massive diffusion models down to just a quarter of their original size without ruining your image quality. The comfyui-starnodes-modelconverter is a custom node package for ComfyUI that introduces advanced, profile-based quantization to intelligently compress heavy models like Flux, SDXL, and Wan. Instead of a basic blanket compression, this project analyzes your model layer by layer and uses smart blacklists to keep highly sensitive elements like embeddings and normalizations in high precision while safely shrinking everything else. It even lets you split all-in-one checkpoints into clean, separate components for easier management.

🆔 @hackernewsgithubprojects

📦 fiigii/ai-comp

ai-comp

The optimizing compiler that solves a complex coding challenge by translating high-level code into highly efficient parallel instructions. ai-comp was built to conquer Anthropic's performance engineering take-home exam, which asks developers to manually optimize a tree-traversal and hashing program running on a simulated virtual machine. Instead of spending hours tweaking the machine code by hand, this project automates the entire process. It takes a human-readable description of the program and intelligently schedules instructions to run simultaneously on multiple virtual execution engines. The coolest part is how it automatically eliminates redundant memory steps, ensuring the simulated processor never wastes a single clock cycle waiting for data that it already has. It...

📰 https://news.ycombinator.com/item?id=48911420

🆔 @hackernewsgithubprojects

📦 smithandrewjohn/kindalive

Kindalive

Model robot emotions through simulated body chemistry instead of simple programming labels. With kindalive, you type a plain text paragraph, and a smart interpreter translates your words into real-time chemical impulses like dopamine, cortisol, and adrenaline. A core engine constantly calculates how these simulated chemicals decay and interact, translating the balance into twelve facial muscle movements. This drives a retro LED dot-matrix face that naturally contorts to show realistic, blended emotions without hard-coded rules. You get a fascinating look at emergent behavior, complete with a responsive web dashboard, plus ready-to-use adapters to bring your physical robot face to life on real hardware.

📰 https://news.ycombinator.com/item?id=48878966

🆔 @hackernewsgithubprojects