📦 tencent/workbuddy-bench

WorkBuddy Bench

Drop your coding agents directly into a simulated office environment to see how they handle realistic corporate workloads. Rather than testing simple code snippets, workbuddy-bench runs your agent inside a local Docker sandbox to tackle complex, multi-step tasks reverse engineered from actual professional roles, like developers, product managers, and security analysts. It evaluates how effectively these agents edit real repositories, fix security vulnerabilities, modify front-end web interfaces, and manipulate messy office spreadsheet files. The system runs the agent, captures its step-by-step behavior, and scores the results automatically. It is the perfect way to pressure-test your artificial intelligence on messy, real-world tasks before letting it loose on your production code.

🆔 @hackernewsgithubprojects

📦 lexsi-labs/circuitkit

CircuitKit

CircuitKit is the mechanistic interpretability framework that lets you directly edit and export task-specialized transformer models. Most AI interpretability tools stop at showing you a colorful graph of which attention heads are active. This clever tool goes much further by finding the exact minimal circuit of neurons driving a specific behavior, letting you prune or steer that pathway, and then exporting a fully functional Hugging Face checkpoint of your newly optimized model. It makes deep model surgery accessible to anyone wanting to patch hallucinations, steer behaviors, or aggressively compress neural networks. Grab this tool and start dissecting your AI models today.

🆔 @hackernewsgithubprojects

📦 decodingai-magazine/building-a-coding-agent-from-scratch-course

building-a-coding-agent-from-scratch-course

Build a fully capable AI coding agent from the ground up without relying on complex, pre-made frameworks that hide all the magic. This repository gives you a free eight-lesson course where you construct a terminal coding assistant called decode in Python. You will go beyond the basic text generator loop to construct the actual harness that controls the model. It guides you through designing strict permission gates, running untrusted code in secure sandboxes, compressing conversation history, and even launching parallel teams of subagents to solve tasks. It is the ultimate hands-on guide to mastering how real-world AI coding tools actually operate.

🆔 @hackernewsgithubprojects

📦 role1776/netmon

Netmon

Your self-hosted network monitor can now roast your internet connection using artificial intelligence. Netmon is a brilliant little Python tool that silently watches your home network, runs hourly speed tests, and scans your local Wi-Fi to count connected devices. Every four hours, it bundles this data into a beautiful custom graph and sends it straight to your Telegram app. But the best part is the commentary. Netmon feeds your actual bandwidth metrics to an AI model, generating hilariously sarcastic, cynical status updates blaming your local network freeloaders or questionable internet provider for any sudden slowdowns. It is the perfect, highly entertaining way to keep tabs on your home connection.

📰 https://news.ycombinator.com/item?id=49012930

🆔 @hackernewsgithubprojects

📦 auto-index/autoindex

AutoIndex

AutoIndex is the automated development tool that uses collaborative AI agents to write customized text-processing programs. Instead of forcing you to manually tweak how your search engine chunks, parses, and indexes giant documents, this system pairs an Analysis Agent with a Code Agent to automatically optimize retrieval. The Analysis Agent studies retrieval failures to figure out where the search engine is slipping up, while the Code Agent writes and test-runs Python programs to fix those exact weak points. It is like having a tireless engineer run experiments on your search index until it is perfect. You should definitely check out AutoIndex on GitHub to see automated code generation tackle search...

🆔 @hackernewsgithubprojects

📦 nuemaan/skewadam

SkewAdam

An open-source optimizer called skewadam can train massive mixture-of-experts artificial intelligence models on a single graphics card by cutting optimizer memory usage by over ninety-seven percent. Traditional training methods spend massive amounts of memory tracking optimizer details uniformly across every parameter. Instead of treating the model like a single giant block, this tool smartly skews its memory budget. It keeps full tracking details for the frequently visited core parts of the model, but drops unnecessary buffers for the sparse expert layers. This simple shift shrinks peak memory demands so dramatically that a huge six-point-eight billion parameter model can train smoothly on a single forty-gigabyte graphics card, delivering top-tier performance without...

📰 https://news.ycombinator.com/item?id=49002895

🆔 @hackernewsgithubprojects

📦 medlartea/refertrack

ReferTrack

Track target objects with robots using simple natural language commands. Instead of getting tangled up in complex visual processing, this clever system first points out the target you describe in words, and then smoothly tracks it over time. It uses special indicator tokens to feed past tracking data right back into the video history, so the robot never loses its train of thought even when things get blocked or the camera shakes. It is incredibly good at helping humanoid and legged robots follow objects in the real world. Give it a look if you want your robots to actually follow directions.

🆔 @hackernewsgithubprojects

📦 eyeline-labs/go-with-the-track

Go-with-the-Track

Go-with-the-Track is the video generation framework that uses point-tracking to perfectly composite reference images and control motion. Instead of treating point-tracks as simple paths across a video, this project anchors those points directly to your reference images. By establishing exact, frame-by-frame coordinate matches, it gives you incredible control over how objects move and blend. You can easily feed it custom animated meshes from Blender or extract precise trajectories from real footage. This makes it incredibly easy to stylize existing videos or composite completely new elements with realistic, physical consistency. It is a brilliant way to bring complex video editing ideas to life.

🆔 @hackernewsgithubprojects

📦 codeswithroh/tastemaker

Tastemaker

Stop letting AI turn your designs into generic purple gradients. You can now use tastemaker to lock down a genuine design system before your coding agent writes a single line of code. Instead of feeding you a boring menu of pre-selected colors, this clever tool analyzes your project's mood to generate a completely fresh, mathematically contrast-compliant color palette on the spot. It reads real pixels from your reference images, builds custom logos, grabs beautiful illustrations, and even wires up smooth animations automatically. Best of all, it remembers your taste profile locally so your next project gets even better.

🆔 @hackernewsgithubprojects

📦 kenlasko/monize

Monize

Take back complete control of your personal finances by self-hosting your own full-featured money manager. Monize is a brilliant replacement for classics like Microsoft Money and Quicken, built entirely using artificial intelligence prompting. It lets you import decades of old financial files and track everything from everyday checking accounts and multi-currency credit cards to stock portfolios with daily market updates. The coolest part is the built-in artificial intelligence assistant that lets you ask natural language questions about your spending habits, running fully local models if you want. It is the ultimate private, modern dashboard for your net worth.

🆔 @hackernewsgithubprojects

📦 mindlab-research/delta-mem

Delta Mem: Giving LLMs True Online Memory

Large language models can now dynamically update their own memory during a live conversation without slowing down or needing expensive retraining. A new project called delta-mem introduces a compact online associative memory state that plugs right into a frozen language model. Instead of bloating the context window with endless text retrieval, it projects incoming information into a low-dimensional space and writes it directly into the active state using delta-rule learning. This allows the model to continuously adapt to new interactions on the fly during live inference. It is a game-changer for building smarter AI agents that actually remember and learn from your conversation.

🆔 @hackernewsgithubprojects

📦 derkbreeze/peot

PEOT

Computers can now slice up a long, chaotic video of someone making breakfast and label every single step automatically, without a human ever teaching them what a fried egg looks like. That is the magic behind peot, an artificial intelligence project that handles unsupervised action segmentation. Instead of relying on rigid, perfect categories, this system maps video frames as flexible, probabilistic clouds. It embraces the messy uncertainty of real-world movements, like pouring milk or stirring coffee. By training on these smart, fuzzy representations, it figures out where one task ends and another begins entirely on its own.

🆔 @hackernewsgithubprojects

📦 ovexro/dockpanel

DockPanel Server Manager

This lightweight self-hosted server manager runs on just nineteen megabytes of random access memory while offering a complete container management platform. The project, called dockpanel, is a modern alternative to traditional control panels, built using Rust and React. It packs a massive suite of features into a tiny footprint, including database management, email services, and a built-in terminal. The most impressive capability is its docker-native setup, which gives you over one hundred and fifty one-click application templates and smart git push-to-deploy options with zero downtime. DockPanel makes managing your own servers incredibly simple and fast without eating up your system resources.

🆔 @hackernewsgithubprojects

📦 nvlabs/fastgen

NVIDIA FastGen

NVIDIA researchers just built a system that lets you train massive ten-billion-parameter generative models much faster by shrinking complex AI architectures. It is called fastgen, and it is a PyTorch-based framework designed to solve the massive computing bottleneck of running and training modern diffusion models. Instead of relying on slow, resource-heavy setups, it packs cutting-edge distillation methods like consistency models and distribution matching to compress giant image and video generators into lightning-fast, highly efficient versions. It is incredibly versatile, working with popular architectures like Stable Diffusion, Flux, and major video models. Grab the code, plug in your datasets, and start generating high-quality media in a fraction of the time.

🆔 @hackernewsgithubprojects

📦 ai-driven-dev/framework

The Framework for AI-Driven Development

Framework is the marketplace hub that makes building software with artificial intelligence highly organized and predictable. Instead of letting an AI assistant loosely guess how to write your code, this tool structures the entire process into modular plugins, specific developer agents, and clear step-by-step skills. It maps out your software's memory, creates coding assertions, and establishes a strict workflow so that automated tools follow precise engineering guidelines. If you want to stop chaotic, unstructured AI output and replace it with reliable, repeatable software construction, this system gives your digital assistants the exact playbook they need. Code smarter with structure.

🆔 @hackernewsgithubprojects

📦 jimmy9704/logicir

LogicIR

Restore blurry and noisy images using the fundamental building blocks of computers instead of heavy, power-hungry neural networks. The open source project logicir reimagines image restoration by using differentiable logic gate networks to clean up corrupted pictures. Usually, processing images requires massive floating-point calculations that drain phone batteries and overheat hardware. This repository replaces those expensive math operations with simple logic gates, making the whole system incredibly lightweight. It even includes a tiny, simplified version specifically designed to run directly on microchips and small hardware devices. It is a fascinating glimpse into how we can make advanced image processing lightning-fast and highly efficient.

🆔 @hackernewsgithubprojects

📦 westlake-agi-lab/budcache

BudCache

Researchers have figured out how to generate high-quality images and videos with diffusion models while skipping up to half of the computation. A new system called budcache introduces a budget-constrained caching method that drastically accelerates generation. Instead of running heavy model calculations at every single denoising step, it automatically searches for the optimal steps to reuse previously cached data without sacrificing visual quality. By matching a smart caching schedule against full-compute trajectories, it delivers stunning results in a fraction of the time. It is a brilliant way to make advanced generative media faster and far more efficient.

🆔 @hackernewsgithubprojects

📦 zibuyu-02/iqa-t1

iqa-t1

The image quality evaluator that uses tool-based visual evidence reasoning to judge pictures like a human expert. Instead of just guessing a quality score or relying on vague descriptions, iqa-t1 actually points to concrete visual evidence using specialized tools to explain its decisions. It solves the issue of biased AI judgments by creating a clear, inspectable reasoning chain you can easily see in action. You can run its local interactive interface to see exactly how it inspects a photo, or experiment with its pre-trained models. Check it out to see how your photos hold up under real visual scrutiny.

🆔 @hackernewsgithubprojects

📦 001tmf/blatant-why

blatant-why

You can now run a complete, multi-agent AI pipeline to design custom proteins and antibodies directly from your command line without paying for expensive enterprise software. A new open-source tool called blatant-why connects Claude Code with structural biology tools, allowing you to design binders and nanobodies using your own hardware, high-performance computing clusters, or cloud services. The project automatically coordinates everything from researching databases to running advanced generation models and screening candidates for structural quality. Once your lab results are ready, you can feed the CSV files back into the system to automatically train and improve the next design round.

📰 https://news.ycombinator.com/item?id=47522328

🆔 @hackernewsgithubprojects

📦 charles-0509/grok-register

grok-register

Automate your account creation and authentication workflows directly from the command line. This Go-based tool handles the entire pipeline by registering accounts, solving turnstile challenges, completing secure OAuth flows, and exporting clean JSON credentials for immediate use. You can easily manage background workers, track real-time logs, configure target limits, and safely update credentials using refresh tokens without manual password inputs. Best of all, it keeps your authentication data organized in a local directory structure while supporting direct uploads to a management API. It is the perfect tool for developers wanting to quickly generate and manage API-ready access keys.

🆔 @hackernewsgithubprojects

📦 bytedoger/awesome-openprice

awesome-openprice

Stop overpaying for AI subscriptions and instantly compare prices across the entire web. This clever platform pulls together real-time pricing for popular tools like Claude, ChatGPT, Gemini, and Cursor from dozens of different online shops and community channels. Instead of manually hunting through forums, group chats, and obscure digital storefronts, you get a clean, unified dashboard where you can filter by price, product type, and warranty status to lock in the absolute best rate. It is like a travel comparison site but built entirely to save you money on artificial intelligence tools. Give it a look and stop wasting cash.

🆔 @hackernewsgithubprojects