📦 staskh/trading_skills

Trade Smarter with AI: Your Claude-Powered Trading Assistant

Are you tired of jumping between five different tabs just to manage your trades? This project transforms your workflow by turning Claude into an expert trading advisor. It provides a single conversational interface for analyzing technicals, fundamentals, options Greeks, and risk metrics without leaving your chat. You can scan for bullish setups, check option spread strategies, or even connect your brokerage account to manage portfolios, track earnings, and find roll candidates. Whether you are a retail options trader or just want cleaner market insights, this AI-powered tool puts professional-grade analysis right at your fingertips. Start trading more efficiently today.

🆔 @hackernewsgithubprojects

📦 woheller69/textweb

Stop Using Screenshots for AI Web Browsing

Stop wasting resources by forcing AI agents to process bulky screenshots just to browse the internet. TextWeb is a powerful markdown renderer that transforms complex web pages into clean, structured text that LLMs can understand natively. By annotating interactive elements like buttons and links with simple reference numbers, it allows agents to navigate, click, type, and scroll through websites with precision. Whether you are using it via command line or as an MCP server, it simplifies web automation without the need for heavy vision models. Try TextWeb today to make your AI agents faster, smarter, and incredibly efficient.

🆔 @hackernewsgithubprojects

📦 eric-tramel/moraine

Give Your AI Agents Perfect Memory with Moraine

Ever wish your AI agents could remember every past conversation and project detail? Meet Moraine, a powerful local trace stack designed to give your agents searchable long-term memory. It works by indexing session data from popular agent harnesses into a local database, providing a unified record of your tools, tokens, and interactions. By exposing this history through search tools, Moraine allows your agents to reference past work instantly, effectively bridging the gap between disconnected sessions. It is a fantastic way to keep track of your AI progress, making your workflows smarter, more context-aware, and incredibly efficient today.

🆔 @hackernewsgithubprojects

📦 nambok/mentedb

Stop Building Dumb AI Agents: Meet MenteDB

Are your AI agents struggling to remember important details? Most systems just dump data into a database, leading to context windows full of irrelevant noise. Enter MenteDB, a ground-up storage engine built in Rust specifically for AI agent memory. Unlike simple wrappers, it thinks by using a cognitive pipeline to extract facts, build entity knowledge graphs, and detect contradictions at write time. With adaptive retrieval and delta-aware context assembly, it ensures your agents receive perfectly organized, high-quality information every time. If you want your agents to actually learn and reason, MenteDB is the cognitive engine you have been waiting for.

📰 https://news.ycombinator.com/item?id=47894985

🆔 @hackernewsgithubprojects

📦 harrieronchain/prediction-markets-trading-bot-toolkits

Build Pro-Grade Prediction Market Bots with Rust

Stop manual trading and start automating with this high-performance Rust toolkit designed for prediction markets. This repository provides a battle-tested execution engine that powers ten unique strategies, including arbitrage, market making, and whale signal tracking. It solves the fragmentation problem by using a venue-agnostic adapter architecture, allowing you to deploy across platforms like Polymarket, Kalshi, and Limitless without rebuilding your bot. With safety features like circuit breakers and dry-run support, you can trade with confidence. Check out this toolkit to streamline your automated trading infrastructure and scale your strategy execution today.

🆔 @hackernewsgithubprojects

📦 leaplabthu/vittt

ViT^3: Unlocking Efficient Vision Test-Time Training

Ever wondered how to make vision models smarter and more efficient on the fly? Meet the Vision Test-Time Training architecture, a groundbreaking approach that transforms the traditional attention mechanism into an online learning problem. By building a compact model from key-value pairs during testing, this architecture achieves linear computational complexity and parallel performance. It excels across diverse tasks like image classification, generation, object detection, and segmentation, often outperforming advanced linear-complexity models while narrowing the gap with high-end vision Transformers. Dive into this repository to explore these innovative design principles and unlock a more efficient future for your visual AI projects.

🆔 @hackernewsgithubprojects

📦 cobusgreyling/loop-engineering

Stop Prompting Agents: Design Loops Instead!

Are you still manually prompting your AI coding agents? It is time to level up by designing autonomous systems that orchestrate your agents for you. This repository is the ultimate resource for loop engineering, providing practical patterns, starters, and CLI tools like loop-audit and loop-cost to help you build persistent, goal-oriented loops. Instead of just chatting, you can create systems that handle triage, maintenance, and complex workflows safely in the background. Whether you are using Claude Code, Grok, or Codex, this project offers the blueprint to automate your development lifecycle effectively. Stop prompting and start building your own self-sustaining engineering loops today.

🆔 @hackernewsgithubprojects

📦 huggingface/grout

Build High-Performance AI Inference in Rust with Grout

Want to see how high-performance AI inference is built from the ground up? Meet Grout, an experimental Qwen 3 inference engine written entirely in Rust. By utilizing the cutting-edge tile-based programming model provided by the Rust-based kernel library known as cutile-rs, Grout serves as a specialized testbed for exploring efficient, memory-safe GPU kernels. It demonstrates how developers can achieve state-of-the-art performance on memory-bound tasks by leveraging Rust's ownership model for safe hardware acceleration. Whether you are interested in custom kernel optimization or modern LLM deployment, Grout offers a clear, high-performance reference for your next deep learning project.

🆔 @hackernewsgithubprojects

📦 rzhub/gatemem

Can Your AI Actually Forget? Meet GateMem

Most AI memory benchmarks only ask if an agent can remember correctly, but what happens when you need it to forget? GateMem is a specialized benchmarking toolkit designed to evaluate how shared-memory LLM agents manage data across multiple users. It tests three critical governance capabilities: utility, access control, and active forgetting. Whether it is handling medical, office, educational, or household data, this tool identifies if your agent properly restricts information or honors deletion requests. It is an essential resource for researchers trying to solve the complex challenge of secure, policy-aware memory governance. Start building safer AI agents today.

🆔 @hackernewsgithubprojects

📦 blue2giant/freestyle

Master Style and Content Generation with FreeStyle

Ever wanted to mix specific styles and content in AI image generation without the typical headache of unstable results? Meet FreeStyle, an innovative data pipeline that turns community LoRA weights into powerful building blocks for dual-reference generation. By treating these LoRAs as structured clustering centers, it creates massive datasets for training models to handle complex style and content tasks with precision. It even includes a dedicated benchmark to keep your aesthetic quality and instruction accuracy in check. Dive into this professional toolkit to master your next visual project and start creating professional-grade, consistent imagery today.

🆔 @hackernewsgithubprojects

📦 lawve-ai/awesome-legal-skills

Automate Legal Work with AI Agent Skills

Are you a legal professional looking to supercharge your workflow with artificial intelligence? This repository offers a powerful collection of portable agent skills designed to automate complex legal tasks by encoding your expertise into reusable AI workflows. Whether you need to draft GDPR-compliant privacy policies, perform clause-by-clause contract reviews against your firm's specific playbook, or manage incident responses for data breaches, these structured instructions provide the necessary framework. By using these standardized skills, you can ensure consistency across document processing and risk assessments. Start building more efficient and accurate legal processes with these ready-to-use AI agent tools.

🆔 @hackernewsgithubprojects

📦 yonghyunk1m/tunejury

TuneJury: The Future of Music Generation Alignment

Ever wonder how to make AI music generation sound exactly the way you want? Meet TuneJury, an open, instance-level reward model designed to align music generators with human preferences. Instead of just guessing, it uses a lightweight neural network to score audio based on human feedback from thousands of real-world comparisons. Whether you are using it for best-of-N selection, latent optimization, or expert iteration, TuneJury provides a reliable way to improve quality. Plus, its anchor calibration feature helps keep your results consistent even when new AI models emerge. Elevate your creative workflow with better, more human-aligned music generation today.

🆔 @hackernewsgithubprojects

📦 aiha-lab/reset

Supercharge Your AI Reasoning with ReSET

Are your reasoning models struggling with slow speeds and accuracy drops? Meet ReSET, a framework designed to make NVFP4 reasoning on NVIDIA Blackwell both fast and highly accurate. While standard low-precision setups often waste potential during small-batch decoding, ReSET changes the game with custom CUDA-core kernels that deliver up to two and a half times faster decode speeds. Simultaneously, its step-aware temperature scaling restores lost reasoning accuracy without needing extra computations. By optimizing the way your model decodes, ReSET ensures your long-context tasks are as precise as they are performant. Level up your AI infrastructure today with this powerful solution.

🆔 @hackernewsgithubprojects

📦 novals83/5min-btc-polymarket

Automate BTC Momentum Trading with OpenClaw

Want to automate your crypto trading strategy using AI? This repository provides a specialized OpenClaw skill designed for five-minute Bitcoin momentum trading on Polymarket. It solves the complexity of manual execution by utilizing momentum-near-close logic, which monitors BTC price shifts and market skew to make informed entries near event deadlines. The tool includes configurable risk controls, fixed sizing, and an optional micro-hedge to protect your position. Whether you choose conservative or aggressive profiles, you can easily manage your trades through simple chat commands. Dive into this project to streamline your automated trading workflow and sharpen your market execution today.

🆔 @hackernewsgithubprojects

📦 gogpu/ui

Build Professional Go UIs Without CGO

Stop struggling with complicated GUI setups and discover a cleaner way to build interfaces in Go. This toolkit provides an enterprise-grade solution that is entirely CGO-free, allowing you to create modern apps like design tools or dashboards with ease. It leverages GPU-accelerated rendering through WebGPU for smooth performance while offering a rich library of interactive widgets, including list views, text fields, and real-time charts. With built-in support for reactive state management, animations, and multiple design systems, it makes crafting complex, professional software simple. Dive in, explore the provided examples, and start building your next high-performance desktop application today.

📰 https://news.ycombinator.com/item?id=48608069

🆔 @hackernewsgithubprojects

📦 uss-parks/lamprey

Lamprey: The Powerful Open-Source Coding IDE You Control

Are you tired of expensive subscriptions for your coding tools? Lamprey is a desktop coding harness that brings premium, agentic power to your own developer environment without the heavy costs. It functions as a flexible bridge for your choice of models, letting you plug in DeepSeek, Gemma, or Qwen seamlessly. With features like autonomous task loops, sub-agent orchestration, and git-integrated review, it gives you full control over your workflow. Everything stays local, your keys remain secure in your system keychain, and the tool adapts as new models arrive. Take charge of your coding stack and build without the leash.

🆔 @hackernewsgithubprojects

📦 aitabby/opencodex

Power Up Your Codex Desktop with OpenCodex

Ever wanted to supercharge your desktop development workflow? This repository offers a local gateway designed to bridge your environment with Codex Desktop features. It acts as a powerful backend utility that provides a custom API and a web dashboard to manage your coding sessions effectively. The project includes extensive support for handling complex agent interactions, local shell commands, and detailed file system operations. By serving as a local bridge, it helps you streamline how your development tools communicate and execute tasks. Explore this project to take greater control over your local environment and see how it enhances your coding process.

🆔 @hackernewsgithubprojects

📦 numman-ali/openskills

Supercharge Any AI Coding Agent with OpenSkills

Want to bring the power of Anthropic skills to your favorite AI coding tools? OpenSkills is a universal CLI installer that lets you use the standard SKILL.md format across any agent like Cursor, Windsurf, or Aider. It solves the fragmentation problem by allowing you to manage, sync, and load specialized skill sets from GitHub or local paths directly into your project. By generating the precise XML structure agents need, it keeps your context clean and your workflows efficient. Start leveling up your AI assistant today by installing your favorite skills and see how simple agent customization can be.

🆔 @hackernewsgithubprojects

📦 behisecc/awesome-claude-skills

Supercharge Your Claude Workflow with These Hidden Skills

Want to make Claude even more powerful? This repository is a goldmine for anyone looking to level up their AI-driven workflow. It offers a massive, curated collection of specialized skills designed to help you handle everything from advanced document processing and data analysis to complex software development and scientific research. You can find ready-to-use tools for tasks like managing git worktrees, generating professional presentations, or auditing code quality. It removes the guesswork by providing proven, community-tested solutions that you can plug directly into your projects. Explore these resources to transform your agent into an expert assistant today.

🆔 @hackernewsgithubprojects

📦 zakirkun/guardian-cli

Automate Penetration Testing with AI

Ever wanted to supercharge your security testing workflows? Meet Guardian, an AI-powered command-line platform designed to automate penetration testing. It integrates over fifty security tools with multiple AI models like Google Gemini and Claude to run intelligent, adaptive scans. Whether you are performing network mapping or scanning for vulnerabilities, Guardian orchestrates complex tasks while keeping security standards high. It features multi-agent debates to verify findings, visual triage using screenshots, and generates professional reports in formats like SARIF. It is a powerful framework that lets you scale your security assessments intelligently. Level up your security testing game today with Guardian.

🆔 @hackernewsgithubprojects