📦 fronalabs/frona

Build Your Own Private Autonomous AI Agents

Ever wish you had a personal AI that actually gets things done on its own? Check out this powerful self-hosted platform that lets you create autonomous agents to handle complex tasks. You can give them tools to browse the web, execute code, build applications, and even make phone calls. The engine is built in Rust for speed and security, featuring isolated sandboxes and credential vaults so your sensitive data remains under your full control. It is perfect for automating workflows by delegating work between specialized agents. Take command of your own private AI infrastructure and automate your daily tasks today.

📰 https://news.ycombinator.com/item?id=47655176

🆔 @hackernewsgithubprojects

📦 pgrwl/pgrwl

Protect Your Data with This Cloud-Native PostgreSQL Tool

Ever worry about losing your critical database transactions in a sudden crash? Meet this powerful PostgreSQL write-ahead log receiver designed for total peace of mind. It acts as a robust, container-friendly alternative that streams logs directly from your database, ensuring zero data loss. The tool handles everything from automatic compression and encryption to seamless uploads to remote storage like S3 or SFTP. With built-in retention policies and simple restore capabilities, it makes disaster recovery feel effortless. Whether you are managing complex Kubernetes clusters or standard servers, this solution keeps your backups durable and secure. Simplify your data protection strategy today.

📰 https://news.ycombinator.com/item?id=47554570

🆔 @hackernewsgithubprojects

📦 gizmo64k/soulplayer-c64

Running a Real AI Transformer on a Commodore 64!

Ever imagined a retro computer running modern artificial intelligence? This incredible project brings a real two-layer transformer, the same technology powering today's most advanced AI models, directly to a Commodore 64. By utilizing clever quantization to pack twenty-five thousand parameters into just twenty-five kilobytes of memory, the developer crafted a model capable of functioning on 1980s hardware. Using hand-written assembly language and fixed-point arithmetic instead of standard multiplication, it generates tokens at a slow, deliberate pace. Explore how this project bridges decades of computing history to show that even an iconic retro machine can possess a digital soul.

📰 https://news.ycombinator.com/item?id=47839645

🆔 @hackernewsgithubprojects

📦 dredozubov/hazmat

Secure Your AI Coding Agents with Hazmat

Want to let your AI coding agents run wild without putting your machine at risk? Meet Hazmat, the essential security layer for AI on macOS. It solves the danger of giving agents full access to your sensitive files and credentials by wrapping them in robust, OS-level containment. With features like dedicated user isolation, kernel-enforced sandboxing, and a built-in firewall, it prevents unauthorized exfiltration while keeping your workflow seamless. Even better, its core security policies are formally verified with TLA to ensure maximum reliability. Install it today to give your AI agents the autonomy they need and the safety you deserve.

📰 https://news.ycombinator.com/item?id=47677488

🆔 @hackernewsgithubprojects

📦 kpolley/redai

Stop Guessing, Start Proving: AI Security with RedAI

Are you tired of security tools that just flag potential issues without proving they are actually exploitable? Meet RedAI, a powerful terminal workbench that transforms how you discover and validate vulnerabilities. Unlike traditional scanners, RedAI uses advanced AI agents to perform live validation against a running application. Whether it is interacting with a web interface in Chrome or running tests in an iOS simulator, these agents provide concrete proof like reproduction scripts and screenshots. By automating the verification process, RedAI helps you distinguish real threats from false positives, giving you actionable evidence to strengthen your security posture today.

📰 https://news.ycombinator.com/item?id=47869732

🆔 @hackernewsgithubprojects

📦 francescostabile/numasec

The AI Agent Transforming Penetration Testing

Stop relying on static scanners that only find isolated issues and start using an intelligent, automated approach to security. This project is a powerful, open-source AI agent designed specifically for penetration testing. It works with any large language model to go beyond simple probing by using thirty-seven native security tools and stateful browser automation to actually chain vulnerabilities together. By maintaining an evidence graph, it turns complex findings into actionable attack paths rather than just piles of disconnected data. Whether you need reconnaissance or full exploitation, this tool provides a smarter, deeper way to secure your applications.

📰 https://news.ycombinator.com/item?id=47674481

🆔 @hackernewsgithubprojects

📦 mcowger/plexus

Stop Rewriting Your AI Code

Are you tired of rewriting your application every time you want to switch AI providers? Meet Plexus, a unified gateway that acts as a single point of entry for top models from OpenAI, Anthropic, and Gemini. Instead of struggling with different integrations, you can swap models instantly without changing a single line of your client code. It handles heavy lifting like OAuth authentication, precise quota tracking, and even specialized tasks like audio transcriptions through an OpenAI-compatible interface. Plexus turns fragmented AI workflows into one seamless system, saving you massive amounts of development time. Give your projects the flexibility they truly deserve today.

📰 https://news.ycombinator.com/item?id=47483643

🆔 @hackernewsgithubprojects

📦 vincentkoc/slacrawl

Take Control of Your Slack Data with This Powerful Terminal Tool

Are you tired of Slack search failing you when you need to find that one specific message from months ago? Enter this powerful terminal tool that puts your workspace data directly into your hands. By mirroring your messages, channels, and user history into a local SQLite database, it lets you perform lightning-fast full-text searches and run custom SQL queries completely offline. Whether you want to back up your history, perform ad-hoc analysis, or just keep your data local, this command-line application makes it happen effortlessly. Give your Slack workflow an upgrade and take full ownership of your data today.

📰 https://news.ycombinator.com/item?id=47304989

🆔 @hackernewsgithubprojects

📦 ybouane/liquidglass

Create Realistic Glass Effects with LiquidGlass

Want to bring stunning, high-end liquid glass effects to your web projects? Meet LiquidGlass, a powerful library that lets you apply realistic glass refraction, blur, and chromatic aberration directly to your HTML elements. By leveraging WebGL shaders, it enables complex visual layering that makes UI elements feel truly tactile. It handles everything from multi-light specular highlights to custom Fresnel reflections with ease. You can even set elements to be dynamic for real-time updates. If you are looking to add that polished, modern aesthetic to your web designs, this library makes implementing professional-grade depth effects incredibly simple and efficient.

📰 https://news.ycombinator.com/item?id=47741405

🆔 @hackernewsgithubprojects

📦 mwallner/mergetopus

Stop Fearing Complex Git Merges

Struggling with massive, messy merge conflicts that grind your team to a halt? Meet Mergetopus, the tool designed to untangle complex repository merges by breaking them down into manageable pieces. Instead of tackling one giant, overwhelming conflict, it splits your work into parallel slice branches that different developers can resolve simultaneously. It automatically creates an integration branch, handles non-conflicting files, and tracks your progress, keeping your history clean and accurate. Whether you are coordinating a large team or managing tricky version cascades, this project makes collaborative merging safer and more efficient. Say goodbye to merge dread and start streamlining your workflow today.

📰 https://news.ycombinator.com/item?id=47748733

🆔 @hackernewsgithubprojects

📦 arknill/claude-code-hidden-problem-analysis

Why Claude Code Is Burning Through Your Tokens

Are your Claude Code sessions hitting rate limits way too fast? This repository offers a deep dive into serious cache bugs that have caused massive token inflation for power users on Max plans. By setting up a transparent proxy to capture thousands of requests, the project uncovered specific technical failures, like silent tool result truncations and incorrect cache prefixes that force expensive reloads. It is an essential resource for understanding how these client side glitches silently waste your budget. Dive into this analysis to learn which versions to avoid and how to optimize your setup for better efficiency.

📰 https://news.ycombinator.com/item?id=47608681

🆔 @hackernewsgithubprojects

📦 openswarm-ai/openswarm

Take Control of Your AI Agent Swarms

Managing multiple AI agents can quickly become a chaotic nightmare when you are stuck toggling between different terminal windows and losing track of progress. Open Swarm solves this by acting as your dedicated mission control center for a swarm of AI agents. Its spatial dashboard allows you to launch, monitor, and coordinate parallel agents across an infinite, zoomable canvas. You get unified approval workflows for tool requests, intelligent message branching for conversation control, and complete visibility over every agent's work. It is the perfect local-first solution to reclaim order, boost your productivity, and effectively scale your AI-powered development workflow.

📰 https://news.ycombinator.com/item?id=47593100

🆔 @hackernewsgithubprojects

📦 sahil-shubham/bhatti

Run AI Agents Safely with Self-Hostable MicroVMs

Are you tired of worrying about the security of your AI coding agents? Meet Bhatti, an open-source orchestrator that lets you run secure, self-hosted microVMs for your agents with ease. Instead of risking your main system, Bhatti creates isolated Linux environments in seconds, providing full kernel and filesystem protection. It is incredibly efficient, allowing you to pause sandboxes to save resources and resume them in mere microseconds for lightning-fast execution. With features like diff snapshots and multi-tenant isolation, it is the perfect sandbox for your development workflow. Give your AI agents a safe, dedicated home today by trying it out.

📰 https://news.ycombinator.com/item?id=47819407

🆔 @hackernewsgithubprojects

📦 varnan-tech/opendirectory

Supercharge Your AI Agents with OpenDirectory

Want to give your AI agents superpowers without teaching them complex tasks from scratch? Meet OpenDirectory, a powerful registry designed for growth automation, technical marketing, and go-to-market workflows. Instead of starting from zero, you can simply download pre-built skills like SEO keyword research tools, LinkedIn post generators, or even automated pull request explainers. It integrates seamlessly with agents like Claude Code, allowing you to instantly install and run these specialized capabilities via your terminal. It is the perfect way to automate your daily grind and boost productivity, so start exploring the catalog and supercharge your AI workflows today.

📰 https://news.ycombinator.com/item?id=47847832

🆔 @hackernewsgithubprojects

📦 zilliztech/vector-graph-rag

Build Faster Graph RAG with Pure Vector Search

Tired of RAG systems that struggle with complex multi-hop questions? This project offers a breakthrough by using pure vector search to perform graph-based reasoning without needing a separate, heavy graph database. It works by encoding entities and relations as vectors in Milvus, enabling efficient subgraph expansion for your data. By replacing slow, iterative agent loops with a single-pass LLM reranking process, it significantly cuts down on both computation and operational costs. It is an ideal tool for anyone looking to achieve state-of-the-art performance on knowledge-intensive tasks, helping you build smarter, faster, and more efficient AI applications today.

📰 https://news.ycombinator.com/item?id=47684290

🆔 @hackernewsgithubprojects

📦 nimaoth/nev

Why You Should Try the Nev Text Editor

If you are tired of switching between terminal and desktop editors, check out Nev. It is a powerful, keyboard-focused text editor written in Nim that runs seamlessly in both your terminal and as a native GUI. By pulling inspiration from popular tools like Neovim and Zed, it brings heavy hitters like Tree-sitter for syntax highlighting, full LSP support, and even WASM plugin capabilities right to your fingertips. Whether you need integrated debugging or built-in Git tools, Nev streamlines your workflow into one fast, efficient environment. Give it a try if you want a modern, unified coding experience.

📰 https://news.ycombinator.com/item?id=47884872

🆔 @hackernewsgithubprojects

📦 facepunch/sbox-public

Is This The Future of Game Development?

Ever wanted to build your own game without wrestling with complex, outdated tools? Meet this powerful open-source game engine that is changing the game for creators everywhere. Built on the advanced technology of Valve's Source 2, it leverages the speed of modern C sharp and a smart, component-based workflow. You can take advantage of lightning-fast hot reloading to see your code changes instantly in the editor, or use visual scripting to craft complex logic without typing a single line. It is a fresh, intuitive sandbox designed to make professional-level game development accessible. Dive in today and start creating your masterpiece.

📰 https://news.ycombinator.com/item?id=47885456

🆔 @hackernewsgithubprojects

📦 highflame-ai/zeroid

Stop Impersonation: Secure Your AI Agents with ZeroID

Ever wonder how to truly identify and govern autonomous AI agents in production? Today, most agents impersonate users through shared service accounts, creating massive security gaps. ZeroID solves this by treating every AI agent as a first-class entity with a unique, cryptographically verifiable identity URI. It enforces delegated authority, allowing agents to perform tasks on behalf of users or other agents while maintaining a clear, auditable delegation chain. With support for real-time revocation and policy-based scope attenuation, it keeps your automated workflows secure and accountable. Stop relying on shared secrets and start building trust into your agentic systems with ZeroID today.

📰 https://news.ycombinator.com/item?id=47674346

🆔 @hackernewsgithubprojects

📦 google-labs-code/design.md

Give Your Coding Agents a Design System with DESIGN.md

Struggling to keep your AI coding agents consistent with your visual brand? Meet the DESIGN.md format, a powerful way to define a persistent design system for your agents. By combining machine-readable tokens with human-readable rationale, this tool ensures your AI understands exactly how to apply your typography, colors, and layout rules. You can even use the included CLI to lint your design files, catch contrast errors, or export configurations directly into Tailwind CSS. It is the perfect bridge for creating high-quality, brand-accurate code automatically. Start using this structured approach today to give your AI agents a professional visual identity.

📰 https://news.ycombinator.com/item?id=47887123

🆔 @hackernewsgithubprojects

📦 droid-vm/droidvm

Run Desktop Linux Directly on Your Android Phone

Have you ever wanted to turn your smartphone into a fully functional desktop computer? DroidVM makes this possible by allowing you to run a virtual machine directly on your Android device. It leverages virtualization technology to create a complete Linux environment right in your pocket without needing a secondary computer. Whether you want to run desktop applications or experiment with new operating systems, this tool bridges the gap between mobile convenience and desktop power. It is an incredibly efficient way to supercharge your device for productivity or development tasks. Try it out to transform your phone into a portable workstation today.

📰 https://news.ycombinator.com/item?id=47887171

🆔 @hackernewsgithubprojects

📦 zimengxiong/winmux

Supercharge Your Mac Workflow with WinMux

Are you tired of managing messy windows on your Mac and want a more efficient way to work? Meet WinMux, an intuitive, batteries-included window manager built on top of Aerospace to help you take full control of your desktop workspace. This tool streamlines your productivity by offering powerful features like tab groups that organize multiple windows into a single footprint and a persistent sidebar for better spatial awareness. Whether you are using its managed tiling mode or simple corner snapping, WinMux provides a cleaner, faster experience that makes navigating your apps effortless. Give your productivity a massive boost today.

📰 https://news.ycombinator.com/item?id=47666590

🆔 @hackernewsgithubprojects