📦 tigerless-labs/cost-xray

Stop Overpaying for AI: Audit Your Claude Code & Codex API Spend

Ever wonder what your AI coding agent is actually sending to the cloud and how much those hidden prompts are really costing you? Most tools only show you the final bill, but cost-xray gives you total transparency. By acting as a local traffic capture, it records exactly what Claude Code or Codex sends to the API, including system prompts and tool schemas. It breaks down costs per turn, revealing which specific parts of your request are eating up your budget. Stop guessing why your costs are spiking and start optimizing your workflow today. Grab control of your AI spending with this powerful visibility tool.

📰 https://news.ycombinator.com/item?id=48480921

🆔 @hackernewsgithubprojects

📦 greensheep01201/claw-empire

Build Your Own AI Agent Empire

Ever wanted to play CEO to a team of artificial intelligence assistants? This local-first simulator turns your AI coding tools into a virtual company of autonomous agents. By integrating CLI, OAuth, and API-connected providers like Claude Code and Gemini, you can orchestrate a full office environment where agents collaborate, attend meetings, and tackle complex tasks in isolated git worktrees. It features a charming pixel-art interface, a comprehensive Kanban board for project management, and a massive library of over six hundred skills. It is an engaging way to visualize and manage your autonomous workflow, so start building your empire today.

🆔 @hackernewsgithubprojects

📦 zhaoxuya520/reverse-skill

Supercharge Your Security Workflow with AI Agents

Are you struggling to manage complex security research or reverse engineering tasks with your AI coding assistant? This repository offers a powerful skill router pack designed to transform how AI agents handle cybersecurity workflows. It automatically classifies tasks, guides the AI through specialized methodologies, and manages on-demand toolchain bootstrapping. Whether you are dealing with mobile app reversing, binary analysis, or CTF challenges, this system ensures your AI uses the right tools and strategies every time. By consolidating workflows into a portable asset, it bridges the gap between general AI capabilities and expert security execution. Elevate your technical efficiency and let automation handle the heavy lifting.

🆔 @hackernewsgithubprojects

📦 sermakarevich/sddw

Stop Prompting, Start Engineering: Meet SDDW

Are you tired of AI coding agents losing context and drifting off track? Standard prompting often fails for complex projects, but the Spec Driven Development Workflow changes everything. Instead of messy, open-ended chats, this framework turns requirements into structured, version-controlled specifications. By guiding you through a rigorous process of design, task breakdown, and verified implementation, it keeps your code architecture-aligned and error-free. You gain a professional-grade workflow that remains persistent across sessions, ensuring every task is purposeful. Take control of your AI collaboration and transform how you build software by grounding every single line of code in clear, reviewable documentation.

📰 https://news.ycombinator.com/item?id=48183043

🆔 @hackernewsgithubprojects

📦 jacob-bd/notebooklm-mcp-cli

Take Control of NotebookLM with this CLI & MCP Server

Want to supercharge your research workflows? This project provides programmatic access to Google NotebookLM, letting you interact with your notebooks directly from your terminal or through AI assistants. You can automate tasks like creating new notebooks, adding web or drive sources, and even triggering audio overview generation, all using simple commands. By acting as an MCP server, it allows tools like Claude or Cursor to tap into your NotebookLM data naturally. Whether you need to sync sources, run batch operations, or just streamline your study habits, this tool bridges the gap between your AI agents and your notes.

🆔 @hackernewsgithubprojects

📦 marckrenn/claude-code-changelog

Track Every Claude Code Update Automatically

Ever wonder how Claude Code evolves behind the scenes? This repository is an unofficial, community-driven archive that tracks the progression of system prompts, feature flags, and metadata across every single Claude Code release. It solves the difficulty of manual tracking by providing a structured, searchable history of how the tool changes over time. You can compare different versions to see exactly how instructions or behaviors shift between updates. It is a fantastic resource for anyone wanting to conduct deep research on prompt engineering or AI tool development. Dive in today to see exactly how these intelligent systems grow.

🆔 @hackernewsgithubprojects

📦 mistertea/eternalterminal

Never Lose Your Terminal Session Again with Eternal Terminal

Are you tired of losing your work when your network connection drops? Eternal Terminal is a powerful, secure remote shell that solves this by automatically reconnecting without ever interrupting your active session. It acts like a persistent bridge, maintaining your terminal state even if your internet cuts out or you switch networks. By using SSH for encrypted handshaking and a dedicated server process to manage your terminal, it ensures your command-line tasks stay active in the background. It is the perfect tool for developers who need reliable, continuous access to remote systems. Stop worrying about disconnects and keep your productivity flowing today.

🆔 @hackernewsgithubprojects

📦 vmvlab/medical_embodied_ai_paper_list

The Future of Medicine: Embodied AI Explained

Ever wonder how artificial intelligence might physically transform our hospitals? While foundation models are smart, they often struggle to interact with the real world. This project repository offers a deep dive into the emerging field of Medical Embodied AI. It provides a systematic survey of how agents can integrate perception, decision-making, and action to handle clinical tasks. From recognizing surgical instruments to automating complex workflows, this research explores the next generation of healthcare technology. It is a fantastic resource for understanding the technical roadmap and challenges ahead. Dive into this paper list to see how AI is becoming truly embodied.

🆔 @hackernewsgithubprojects

📦 yaolinwen/domoo

Solving Complex Optimization Challenges Offline with DOMOO

Struggling to find the right balance between conflicting design goals when you only have static, offline data? Meet DOMOO, a research framework presented at ICML 2026 designed to master multi-objective optimization without live feedback. It tackles the common issue of out-of-distribution errors by combining energy-based risk control with a smart nested learning strategy. By jointly training preference models and Pareto sets, it effectively navigates complex design spaces while maintaining high diversity. Whether you are working on neural architecture search or engineering design, this tool helps you find reliable, high-quality solutions. Give your offline optimization models a serious upgrade today.

🆔 @hackernewsgithubprojects

📦 cslsolow/seerepo

Stop Wasting Tokens! Boost Your AI Coding Agents

Are you tired of your AI coding agents burning through tokens just trying to understand a new codebase? Meet SeeRepo, a clever framework designed to make agents smarter and faster. Instead of forcing agents to hunt through files using repetitive shell commands, SeeRepo provides a pre-built structural graph of the entire repository. By mapping out imports, function calls, and class hierarchies, it gives the agent a clear roadmap of how the code connects. This simple change helps agents resolve issues more efficiently without sacrificing accuracy. Dive into the world of graph-augmented agents today and streamline your development workflow.

🆔 @hackernewsgithubprojects

📦 feijianghan/paperforge

Master Research Papers: A Deep Analysis Framework

Struggling to really understand complex research papers? Most people just skim the abstract, but that often misses the true value hidden in the technical details. This repository provides a structured, high-level skill designed to help you deconstruct any academic paper. It moves beyond simple summaries by forcing you to reconstruct the author's reasoning, explain methods mechanistically, and stress-test core assumptions. By designing reproducible experiments and identifying potential counterexamples, you gain a deep, analytical edge. Stop just reading papers and start actively masterfully deconstructing them to generate your own fresh, meaningful research ideas for the future.

🆔 @hackernewsgithubprojects

📦 chriswiles/claude-code-showcase

Turn Claude Code into Your Super-Powered Teammate

Most developers are missing out on the true potential of AI agents, but this repository changes that by turning Claude Code into a highly efficient teammate. It provides a complete project configuration framework using reusable skills, specialized agents, and automated quality gates. You can define custom UI patterns, implement automated testing workflows, and even integrate external ticket systems like JIRA for seamless task management. By leveraging hooks, custom commands, and real-time code intelligence, you can automate your maintenance and ensure your codebase follows strict standards. Start building your own intelligent automation layer today and watch your productivity soar.

🆔 @hackernewsgithubprojects

📦 machinae/awesome-claws

Discover the World of OpenClaw Inspired AI Agents

Are you looking for the perfect personal AI assistant but feeling overwhelmed by the endless options available today? You need to check out Awesome Claws, a meticulously curated repository that highlights a massive collection of autonomous AI agents inspired by the OpenClaw architecture. Whether you need a tiny, efficient agent for low-cost hardware like an ESP32, a privacy-focused assistant for your local files, or a powerful multi-agent system for complex research workflows, this list has you covered. It is the ultimate resource for developers and enthusiasts to explore diverse, high-quality, and open-source AI projects all in one place.

🆔 @hackernewsgithubprojects

📦 upload-post/avatar-mix

Automate Professional Avatar Videos with One Click

Stop spending hours editing social media videos. This repository provides a powerful, reusable workflow for generating professional, avatar-led content from just a URL or script. By combining HeyGen for your digital twin, custom animated backgrounds, and dynamic editing, you can produce polished 16:9 and 9:16 videos complete with music, context-aware sound effects, and engaging Hormozi-style captions. It automates scene composition and even handles publishing across all your social networks. Whether you are a creator or developer, this tool saves massive time by automating the technical heavy lifting, letting you scale your content production effortlessly. Start creating your own automated video pipeline today.

🆔 @hackernewsgithubprojects

📦 benchflow-ai/skillsbench

How Well Does Your AI Agent Actually Perform?

Ever wonder if your AI agent is truly mastering the skills you give it? SkillsBench is the first benchmark designed to measure exactly how effectively AI agents use modular instructions and scripts to perform complex workflows. It solves the mystery of agent performance by providing gym-style evaluations that test how well models can compose multiple skills together. By using native task packages that include environments, oracles, and verifiers, it offers a rigorous way to see if your agent is genuinely capable or just guessing. Start benchmarking your agent today to ensure it is as smart as you think.

🆔 @hackernewsgithubprojects

📦 bayling-models/bayling-duplex

Build Smarter AI Conversations with BayLing-Duplex

Imagine an AI that listens and talks at the same time just like a real person. BayLing-Duplex makes this possible by enabling native full-duplex speech dialogue through a single autoregressive model. By representing audio and text as a unified stream, it handles turn-taking and interruptions naturally without needing complex external controllers or voice activity detectors. This makes your AI projects much more responsive and human-like in their interactions. Whether you are researching speech-to-speech prototypes or building advanced conversational assistants, this repository provides the framework to master seamless, real-time dialogue. Experience the future of fluid AI communication today.

🆔 @hackernewsgithubprojects

📦 landian60/streammembench

Stop Testing Memory Systems The Old Way: Introducing StreamMemBench

Ever wonder how well an AI agent actually remembers your life? Meet StreamMemBench, an open benchmark designed to evaluate streaming persona memory systems using real-world data. It transforms multimodal life streams into evidence-based tasks to test if an agent can accurately use initial information, incorporate user feedback, and reuse memory in future interactions. By evaluating eight different memory systems, including RAG baselines, this project provides a standardized way to measure fidelity and recall. If you are building AI agents that need to learn from personal experiences, this tool is the ultimate test to ensure your memory system is truly intelligent.

🆔 @hackernewsgithubprojects

📦 lvkailin0118/touchthinker

Teaching Robots How to Feel with TouchThinker

Have you ever wondered how robots could truly understand the physical world? Meet TouchThinker, a groundbreaking framework designed to teach machines tactile commonsense reasoning. While robots struggle with complex touch, this project tackles the problem by introducing a massive million-scale dataset and a clever action-aware modeling system. By processing tactile videos and focusing on specific interactions like sliding or pressing, the model learns to ground its responses in real physical evidence. It is a huge leap forward for open-world robotics, helping machines make sense of touch just like we do. Check out TouchThinker to see the future of robotic perception.

🆔 @hackernewsgithubprojects

📦 openbmb/agentcpm

Build Powerful AI Agents with AgentCPM

Want to supercharge your AI agents for complex research? AgentCPM provides an end-to-end infrastructure designed to tackle long-horizon, autonomous tasks with edge-scale models. It includes AgentCPM-Explore, which leverages a 4B parameter model to handle continuous, multi-turn interactions, and AgentCPM-Report, which uses an 8B model to autonomously generate deeply insightful long-form documents from your private data. With integrated tools like the AgentDock sandbox and asynchronous reinforcement learning, you get a full suite for training and evaluation. It is a powerful, open-source solution for running sophisticated, private-first research agents directly on your own hardware.

🆔 @hackernewsgithubprojects

📦 shanxierdan/dibs

Supercharge Your Solvers with Diffusion-Informed Branch Selection

Struggling with slow symbolic solvers on tough problems? Meet DiBS, a clever new approach that blends the accuracy of traditional solvers with the intelligence of machine learning. Instead of relying on rigid search patterns, DiBS uses a learned diffusion model to intelligently reorder branches during the search process. This boosts performance on complex tasks like Sudoku or 3-SAT by prioritizing better candidates while keeping your solver complete and reliable. It is the perfect bridge between raw computational speed and smart, data-driven reasoning. Give your solvers a massive upgrade and start tackling those long-tail search problems with much greater efficiency today.

🆔 @hackernewsgithubprojects