📦 rivet-dev/agent-os

Build AI Agents Faster with Agent OS

Imagine giving your AI agents a dedicated operating system designed specifically for speed and efficiency. Agent OS is a game-changing framework that replaces slow, heavy virtual machines with lightweight WebAssembly and V8 isolates. By cutting cold starts down to just six milliseconds, this tool makes deploying intelligent agents thirty-two times cheaper than traditional sandboxes. It handles the complex infrastructure for you, letting you focus entirely on building smarter, more responsive automation workflows without the overhead. Whether you are scaling complex tasks or building real-time interactive apps, Agent OS provides the high-performance foundation you need to truly level up your development.

📰 https://news.ycombinator.com/item?id=48023247

🆔 @hackernewsgithubprojects

📦 hmbown/deepseek-tui

Supercharge Your Terminal with DeepSeek TUI

Want to turn your terminal into a powerful AI-driven workspace? Meet DeepSeek TUI, a lightning-fast coding agent that brings the intelligence of DeepSeek models directly into your command line interface. It solves the frustration of switching windows by letting you interact with your code, run shell commands, and browse the web without ever leaving your editor. With support for massive context windows, real-time chain-of-thought reasoning, and handy features like session resume and workspace snapshots, it is built to keep your workflow fluid and productive. Elevate your development experience and let this intelligent agent handle the heavy lifting for you.

🆔 @hackernewsgithubprojects

📦 vectifyai/pageindex

Stop Using Vectors: Meet PageIndex

Are you tired of inaccurate search results when digging through long, complex documents? Traditional vector-based methods often miss the mark because they focus on similarity rather than actual relevance. Enter PageIndex, a revolutionary open-source framework that ditches vector databases and chunking in favor of reasoning-based retrieval. It builds a hierarchical, tree-like structure from your documents, allowing AI models to think and navigate information just like a human expert would. By using this smarter, context-aware approach, you get better explainability and precision for your technical or legal files. Give your AI a better way to read today.

📰 https://news.ycombinator.com/item?id=48023942

🆔 @hackernewsgithubprojects

📦 flowexec/flow

Automate Your Workflow Locally with Flow

Are you tired of relying on clunky cloud tools just to automate your daily development tasks? Meet Flow, an incredible local-first automation platform that brings power and privacy back to your own machine. It streamlines your developer life by managing complex workflows, handling sensitive secrets, and using custom templates entirely on your local infrastructure. Because everything runs offline, you gain lightning-fast execution and total security for your data. It is the perfect way to eliminate repetitive manual steps while keeping your environment clean and fast. Try integrating Flow into your pipeline today to reclaim control over your entire development process.

📰 https://news.ycombinator.com/item?id=48023887

🆔 @hackernewsgithubprojects

📦 jal-co/shieldcn

Level Up Your README With Beautiful shadcn-Style Badges

Want your GitHub README to look as sleek as a modern web app? Meet the repository that brings the clean, minimalist aesthetic of shadcn-ui directly to your project badges. While standard badges can feel outdated, this tool renders them as actual button components using Satori, ensuring your project status, versioning, and stats match your modern design system perfectly. It supports dozens of providers, custom SVG icons, and even integrates with AI coding agents to automate your badge updates. Upgrade your documentation to look professional and cohesive without compromising on functionality or design. Give your project the premium look it truly deserves.

📰 https://news.ycombinator.com/item?id=47907232

🆔 @hackernewsgithubprojects

📦 kubernetes-sigs/kubebuilder-declarative-pattern

Build Kubernetes Operators Faster

Stop fighting with complex Kubernetes controller logic and start managing your infrastructure with declarative patterns. This toolkit lets you build powerful operators by simply defining the desired state of your resources rather than writing endless imperative code. It automates the heavy lifting by handling reconciliation and status reporting for you, which significantly slashes your development time. You can easily manage multi-resource deployments through simple YAML configurations, making your workflows cleaner and far more predictable. It is a game changer for anyone looking to simplify their automation strategy. Dive into this approach today and master your clusters with ease.

📰 https://news.ycombinator.com/item?id=48025444

🆔 @hackernewsgithubprojects

📦 lechatp/rootasrole

Stop Using sudo: A Better Way to Manage Privileges

Are you tired of giving users full root access just to run a single command? Most systems ignore the principle of least privilege, but there is a smarter way to handle security. This project acts as a high-performance, memory-safe alternative to standard privilege delegation tools. Instead of handing over total control, it lets you assign specific, fine-grained roles and tasks to users. You can define thousands of complex, rule-based policies that remain incredibly fast at scale, even during automation tasks. Start enforcing better security today by switching to a more precise, role-based approach for your Linux system administration.

📰 https://news.ycombinator.com/item?id=48025639

🆔 @hackernewsgithubprojects

📦 lsferreira42/nfsdiag

Stop struggling with NFS issues!

Are you tired of constantly battling frustrating network file system errors that bring your workflow to a sudden halt? Meet the perfect diagnostic companion designed to act as a specialized doctor for your network storage environment. This tool automatically probes your configuration to identify connectivity bottlenecks and permission conflicts that typically cause slow performance. By performing health checks and verifying mount points directly, it replaces manual troubleshooting with clear, actionable insights. Whether you are managing complex servers or local setups, this application makes maintenance effortless. Give your infrastructure the checkup it deserves and get back to work faster than ever.

📰 https://news.ycombinator.com/item?id=47985958

🆔 @hackernewsgithubprojects

📦 pysimplegui/pysimplegui

Build Python GUIs Effortlessly with PySimpleGUI

Tired of wrestling with complex code just to build a simple window for your Python projects? Meet PySimpleGUI, the ultimate tool designed to make creating graphical user interfaces actually fun. Instead of struggling with the intricate details of libraries like Tkinter, Qt, or WxPython, this library abstracts away the heavy lifting into a simple, intuitive syntax. Whether you are a hobbyist just starting out or a seasoned developer looking to speed up your workflow, it transforms how you build apps. Give your scripts a professional look and feel today and start building amazing interfaces with absolute ease.

📰 https://news.ycombinator.com/item?id=48026412

🆔 @hackernewsgithubprojects

📦 noonghunna/club-3090

Run LLMs on RTX 3090s Like a Pro

Struggling to get high-performance AI running on your home hardware? This community project provides the ultimate toolkit for serving large language models specifically on NVIDIA RTX 3090 graphics cards. By offering optimized, model-agnostic recipes for multiple engines like vLLM and llama.cpp, it solves the headache of complex configurations and performance bottlenecks. You get access to validated setups for single or dual-card builds, including specialized configs for models like Qwen. Whether you need maximum throughput or rock-solid stability for long-context tasks, these verified scripts simplify your workflow. Start building your own powerful local AI infrastructure today with total confidence.

📰 https://news.ycombinator.com/item?id=48026703

🆔 @hackernewsgithubprojects

📦 0xchamin/mcptube

Turn YouTube Into Your Personal Knowledge Base

Ever wish you could instantly turn massive YouTube video libraries into a searchable, intelligent knowledge base. This project makes that a reality by functioning as a specialized server that integrates directly with tools like Claude. It automatically fetches video transcripts and utilizes vision analysis to process visual data, allowing you to ask complex questions and retrieve precise information instantly. By acting as an agentic bridge, it transforms passive watching into active learning without the manual effort. If you want to supercharge your research and stop wasting hours scrubbing through long clips, this clever integration is exactly what you need.

📰 https://news.ycombinator.com/item?id=47754559

🆔 @hackernewsgithubprojects

📦 youssofal/mtplx

Turbocharge Your LLM Inference on Apple Silicon with MTPLX

Tired of waiting for your local AI models to generate text? Meet MTPLX, a powerful native runtime built for Apple Silicon that uses Multi-Token Prediction to supercharge your inference speed by over two times. Unlike other tools that rely on bulky external drafters, MTPLX cleverly utilizes the model’s own built-in heads to predict tokens, ensuring math-correct sampling even at higher temperatures. It offers a seamless, OpenAI and Anthropic-compatible API, making it easy to integrate into your favorite developer tools. Experience faster, more efficient local AI chats today by installing it and unlocking your machine’s full performance potential.

📰 https://news.ycombinator.com/item?id=48016969

🆔 @hackernewsgithubprojects

📦 microsoft/trellis.2

Turn Any Image into High-Fidelity 3D with TRELLIS.2

Ever wanted to turn a single flat image into a high-quality 3D asset in just seconds? Meet the repository that is changing the game for generative 3D. This powerful tool uses a unique sparse voxel structure called O-Voxel to bypass messy, lossy conversions, letting you create complex geometry with full PBR materials directly from images. Whether you need open surfaces or intricate internal structures, this four billion parameter model handles it all with incredible efficiency and professional detail. It is a massive leap forward for anyone looking to streamline their 3D workflow. Dive in and start building the future today.

🆔 @hackernewsgithubprojects

📦 facebookresearch/programbench

Can AI Rebuild Your Code? Meet ProgramBench

Ever wonder if AI can truly understand and reconstruct complex software from scratch? ProgramBench is a groundbreaking research framework designed to test exactly how well large language models handle real world coding challenges. Instead of simple snippets, it evaluates models on their ability to rebuild entire software projects by navigating intricate file structures and dependencies. By automating the evaluation process, it helps developers pinpoint where models struggle with logic and architecture, ultimately pushing the boundaries of AI assisted programming. Whether you are a researcher or a developer, this project offers a vital glimpse into the future of automated software engineering.

📰 https://news.ycombinator.com/item?id=48025412

🆔 @hackernewsgithubprojects

📦 du82/nonograph

Take Control of Your Privacy with Nonograph

Want to publish content online without ever sacrificing your privacy? Meet Nonograph, a lightweight self-hosted platform designed for total anonymity. It eliminates accounts, tracking, and data collection, allowing you to write and share instantly with peace of mind. The project uses Rust for performance and even bundles Tor support within its Docker containers to ensure your identity remains hidden. Whether you want to format posts using Markdown with spoiler support or host it on a tiny server using minimal resources, this tool has you covered. It is the ultimate way to reclaim your digital voice anonymously.

📰 https://news.ycombinator.com/item?id=47912751

🆔 @hackernewsgithubprojects

📦 nrvim/garmin-givemydata

Take Control of Your Garmin Fitness Data

Stop letting your hard-earned fitness stats live trapped inside the Garmin cloud forever. This clever tool finally gives you the power to reclaim your personal health information by pulling your Garmin Connect history directly into a local SQLite database on your own machine. By converting your activity metrics into a structured format, it makes your workout data fully prepared for analysis or even connection to AI models as an MCP server. It is a fantastic way to secure your privacy and unlock advanced insights into your performance. Go ahead and take back ownership of your data today.

📰 https://news.ycombinator.com/item?id=47661527

🆔 @hackernewsgithubprojects

📦 open-gitagent/gitagent-protocol

Stop Managing AI Agents: The Power of GitAgent

Are you tired of juggling fragmented AI agent setups? Imagine turning any repository into a portable, framework-agnostic AI agent just by dropping in a few files. This protocol makes your AI agents truly git-native, meaning you get version control, branching, and collaboration for free. You can define agent identities in simple manifests, enforce strict segregation of duties for compliance, and use standardized tools across different AI models. By treating your agent as a standard repository, you gain the power of familiar workflows like git diff and pull requests. Take control of your AI infrastructure and start standardizing today.

📰 https://news.ycombinator.com/item?id=48033610

🆔 @hackernewsgithubprojects

📦 teng-lin/notebooklm-py

Unlock Google NotebookLM with Python Automation

Ever wish you could automate your research or generate study materials directly from your code? This powerful unofficial tool unlocks the full potential of Google NotebookLM by providing a complete Python API and command-line interface. You can now programmatically batch download your audio overviews, export quizzes as structured data, or even integrate NotebookLM directly into AI agents like Claude Code. It solves the limitation of the standard web interface by giving you total control over your notebooks and data. Unlock new levels of productivity in your research and content workflows by bringing professional automation to your fingertips today.

📰 https://news.ycombinator.com/item?id=48034401

🆔 @hackernewsgithubprojects

📦 vercel-labs/open-agents

Build Your Own Cloud Coding Agents

Ever wish your coding agent could keep working even after you close your laptop? This open source project provides a powerful template for building background agents that run entirely in the cloud. It features a smart three-layer architecture that separates the agent runtime from the execution sandbox, ensuring your coding tasks persist independently. You get full control over file management, shell commands, and even automatic GitHub pull requests. By leveraging isolated sandboxes, it offers a robust way to develop complex AI workflows that handle everything from prompt to deployment. Dive into this repository to start automating your own developer lifecycle today.

📰 https://news.ycombinator.com/item?id=48034362

🆔 @hackernewsgithubprojects

📦 semos-labs/attyx

Attyx: The Fast GPU-Accelerated Terminal Built With Zig

Are you tired of sluggish terminals that require endless configuration files just to feel productive? Meet Attyx, an incredibly fast and lightweight terminal emulator written from scratch in Zig. It offloads rendering to your GPU, delivering buttery smooth performance while packing powerful features like native window splits, tabs, and a built-in status bar directly out of the box. By combining high-speed execution with essential modern workflows, Attyx replaces the need for complex tool chains. If you want a programmable, modern terminal experience that stays out of your way and handles your daily tasks with ease, give this project a look.

📰 https://news.ycombinator.com/item?id=48034880

🆔 @hackernewsgithubprojects

📦 vectifyai/openkb

Stop Building RAG! Use This Wiki-Based Knowledge Engine Instead

Are you tired of traditional AI search systems that rediscover the same information from scratch every time you ask a question? Meet OpenKB, a revolutionary open-source project that flips the script by compiling your raw documents into a structured, living wiki. Instead of using complex vector databases, it uses intelligent agent-based reasoning to cross-link concepts, summarize papers, and keep your knowledge base perfectly in sync. It even handles massive files using hierarchical indexing and outputs clean markdown files compatible with tools like Obsidian. Start building a persistent second brain today and watch your collective knowledge actually compound over time.

📰 https://news.ycombinator.com/item?id=48035324

🆔 @hackernewsgithubprojects