Projects

### Overview: Agentshire is an OpenClaw/QClaw plugin that transforms AI agents into living NPCs within a 3D game town. This plugin combines a dynamic simulation layer with user-generated content (UGC) tools, allowing for a fully immersive experience where agents can interact and collaborate with real-world NPCs. ### Key Features: - **Town Mode**: A low-poly 3D town where agents can live, work, collaborate, and celebrate in real-time. - **Chat Mode**: An IM-style chat interface with agent lists, message history, and multimodal support. - **Dual-Mode Interface**: One-click switch between Town and Chat modes, with quick navigation menus. - **Bilingual UI**: Full Chinese and English interface, auto-detected or manually switchable. - **Core Features**: - **Agent = NPC**: Every sub-agent becomes a 3D citizen with a name, appearance, and personality. - **Cinematic Workflow**: Summon → Rally → Assign → Enter Office → Code → Celebrate → Return to Town, with fully choreographed animations. - **Real-Time Dialog Bubbles**: AI responses appear above NPC heads with a typewriter effect, streaming updates, and pagination. - **Multi-Agent Collaboration**: The steward automatically decomposes tasks into parallel steps with file boundary validation. - **Zero Configuration**: Install the plugin, start the Gateway, and the town runs automatically. - **Town Life**: - **Day/Night Cycle**: 24-hour clock with real-time lighting changes and automatic street/window lights. -

Overview: The gdxAI project is an AI framework for game development, built on top of libGDX. It offers a variety of AI techniques, including steering behaviors, formation motion, pathfinding, behavior trees, and finite state machines. Key features: - Steering Behaviors - Formation Motion - Pathfinding (A*, Hierarchical Pathfinding, Path Smoothing, Interruptible Pathfinding) - Decision Making (State Machine, Behavior Trees) - Infrastructure (Message Handling, Scheduling) Tech stack: Java Why it matters: The gdxAI framework provides essential AI tools for game developers, enabling them to create more realistic and dynamic game environments efficiently. Repository: libgdx/gdx-ai GitHub stars: 1306 GitHub description: Artificial Intelligence framework for games based on libGDX or not. Features: Steering Behaviors, Formation Motion, Pathfinding, Behavior Trees and Finite State Machines GitHub image: 

### Overview: GoLive is an open-source Agent Skill designed to automate the deployment and verification of agent-built products. It simplifies the process of setting up accounts, hosting, databases, domains, email, and payments, allowing developers to focus on building their applications. ### Key Features: - **Automatic Verification**: GoLive detects, plans, approves, applies, and verifies changes, ensuring everything works as expected. - **No GoLive Account Needed**: Supports hosting, databases, domains, email, and payments without requiring a GoLive account. - **Zero-Dependency Node CLI**: Provides a simple CLI for detecting, planning, approving, applying, and verifying changes. ### Tech Stack: TypeScript ### Why It Matters: GoLive streamlines the deployment and verification process, making it easier for developers to focus on building their applications rather than worrying about the infrastructure. It eliminates the need for GoLive accounts, backend hosting, or telemetry, making it a practical solution for anyone looking to take their agent-built products live.

Overview: SmythOS/sre is an open-source, cloud-native runtime environment for building and managing intelligent agents. It offers a secure, modular, and production-ready foundation for AI applications, enabling developers to write, scale, and manage agents across local, cloud, and edge environments. Key features: - Secure, modular, and production-ready - Unified API for AI resources - Built-in security, observability, and 40+ production-ready components - Open-source infrastructure for reliable agent engineering Tech stack: TypeScript Why it matters: SmythOS/sre provides a robust and scalable foundation for AI agent orchestration and lifecycle management, ensuring every builder has the tools to act, not just imagine. It simplifies the process of building, scaling, and managing intelligent agents, making it easier for developers to ship production-ready AI applications.

### Overview: 《经济研究》、《管理世界》、《中国社会科学》、《中国工业经济》、《世界经济》和《经济学(季刊)》等主流经济学与管理类期刊的Claude Code/Codex技能包，涵盖选题、识别策略、表格规范、审稿回复等全流程指导，助您快速发表论文。 ### Key features: - 《经济研究》 - 《管理世界》 - 《中国社会科学》 - 《中国工业经济》 - 《世界经济》 - 《经济学(季刊)》 ### Tech stack: Stata ### Why it matters: 覆盖主流经济学与管理类期刊的Claude Code/Codex技能包，从选题、识别策略到表格规范与审稿回复全流程，助您快速发论文。

Overview: AI Builder Club's skills plugin marketplace offers a comprehensive set of tools for building loop engineers, designed to automate and optimize the work of agents triggered by their own actions. Key features: - Codebase harness: makes any repository agent-ready, including isolated cloud boxes. - Loops: spins up compounding agent loops on a shared, file-based knowledge base. - Agent-context-audit: audits agent context against Claude 5 context-engineering guidance. - Open agent teams: delegates tasks to any CLI agent in a detached tmux session. - SEO growth: playbook for cold start and existing site SEO efforts. - Visual flow GIF: turns articles, workflows, or architectures into static PNGs and animated GIFs. - Karaoke captions: burn word-level highlight captions into videos. Tech stack: Python. Why it matters: This plugin provides a productized version of the setup used by the AI Builder Club, streamlining the process of designing and running loop engineers that automate and optimize agent tasks. Repository: AI-Builder-Club/skills Primary language: Python GitHub stars: 1298 GitHub description: Codebase harness + loop engineer Repository: AI-Builder-Club/skills Primary language: Python GitHub stars: 1298 GitHub description: Codebase harness + loop engineer

### Overview: Mish: A Self Regularized Non-Monotonic Neural Activation Function is an open-source project that introduces a novel activation function called Mish, which is designed to improve the performance of neural networks. This function is particularly effective in reducing the vanishing gradient problem and increasing the model's capacity. ### Key Features: - **Self Regularization**: Mish function is self-regularized, meaning it can handle the vanishing gradient problem without the need for additional regularization. - **Non-Monotonic Behavior**: It exhibits non-monotonic behavior, which can help in improving the model's performance by avoiding the vanishing gradient problem. - **Efficient Implementation**: The function is implemented in Jupyter Notebook, making it easy to understand and experiment with. ### Tech Stack: - **Jupyter Notebook**: The project is primarily developed using Jupyter Notebook, which is ideal for interactive data analysis and machine learning experiments. ### Why It Matters: Mish is a significant contribution to the field of neural network activation functions, offering a faster and more effective alternative to traditional activation functions. Its ability to handle the vanishing gradient problem and its non-monotonic behavior make it a valuable tool for improving the performance of deep learning models. ### Repository: The project is hosted on GitHub at [digantamisra98/Mish](https://github.com/digantamisra98/Mish). It has been starred by 1,300 users, indicating its popularity and utility in the machine learning community. ### GitHub Stars: The repository has been starred by 1

Overview: Puppyone is a context drive solution designed for AI agents, offering Git version control and file-level scoped permissions. Key features: - Context hosting for AI agents - Git version control - File-level scoped permissions - CLI, MCP, Git, bash/SSH, and web app interactions through Access Points Tech stack: Python Why it matters: Puppyone provides a robust and secure environment for AI agents, enabling seamless integration with Git and other tools, making it ideal for developers working on AI projects. Repository: puppyone-ai/puppyone-cloud GitHub stars: 1300 GitHub description: Context drive for your AI agents README excerpt: <div align="center"> <img src="assets/puppyone.svg" alt="Puppyone Logo" width="72" height="72" /> <h1>Puppyone</h1> <p><b>Git-native Context Drive for AI agents.</b></p> <p>Puppyone provides context hosting for AI agents, with Git version control and file-level scoped permissions for every agent.</p> <p> <a href="https://www.puppyone.ai"><img src="https://img.shields.io/badge/Website-puppyone.ai-39BC66?style=flat-square" alt="Website" /></a> <a href="https://www.puppyone.ai/doc"><img src="https://img.shields.io/badge/Docs-Read-D7F3FF?style

### Overview: AI Basketball Analysis is an AI-powered application that uses machine learning to analyze basketball shots and shooting poses. It leverages OpenPose for body keypoint detection and object detection to provide detailed analysis of basketball performances. ### Key Features: - **Upload Videos**: Users can upload basketball videos for analysis. - **POST Requests**: Submit POST requests to an API for more advanced analysis. - **Detailed Analysis**: Results include shot and pose analysis based on object detection data. ### Tech Stack: - Python - OpenPose ### Why It Matters: This project demonstrates how AI can automate and enhance basketball analysis, providing valuable insights into player performance and shot accuracy.

### Overview: Therapeutics Data Commons (TDC) is a collaborative initiative aimed at advancing artificial intelligence (AI) in therapeutic science. It provides a comprehensive platform for researchers to access, evaluate, and develop AI methods for drug discovery and development. TDC offers a variety of resources, including AI-solvable tasks, AI-ready datasets, and curated benchmarks, serving as a bridge between biochemical and AI scientists. ### Key Features: - **Multimodal and Contextualized Machine Learning Foundations for Therapeutics**: Supports AI methods across different therapeutic modalities and stages. - **Curated Datasets and Benchmarks**: Provides AI-ready datasets and benchmarks for drug discovery and development. - **Algorithmic and Scientific Advances**: Facilitates algorithmic and scientific advancements in machine learning for drug discovery. - **Accelerates Method Development and Validation**: Helps in the rapid development, validation, and transition of machine learning methods into biomedical and clinical applications. - **Open-Science Initiative**: Encourages community contributions and collaboration. ### Tech Stack: Jupyter Notebook ### Why It Matters: Therapeutics Data Commons (TDC) is crucial for researchers across various disciplines, providing a unified platform to access and evaluate AI methods for drug discovery. It serves as a bridge between biochemical and AI scientists, facilitating algorithmic and scientific advancements, and accelerating the transition of AI methods into practical applications in therapeutic science. ### Repository: Repository: mims-harvard/TDC Primary Language: Jupyter Notebook GitHub Stars: 1298 ### GitHub Description: Therapeutics Data Commons (TDC):

Overview: xmnlp 是一款开箱即用的中文自然语言处理工具包，提供包括中文词法分析、中文拼写检查、文本摘要、情感分析、文本转拼音、汉字偏旁部首提取、句子表征及相似度计算等多种功能。 Key features: - 中文词法分析 (RoBERTa + CRF finetune) - 中文拼写检查 (Detector + Corrector SpellCheck) - 文本摘要 & 关键词提取 (Textrank) - 情感分析 (RoBERTa finetune) - 文本转拼音 (Trie) - 汉字偏旁部首 (HashMap) - [句子表征及相似度计算](https://mp.weixin.qq.com/s/DFUXUnlH_5BlxwyQYeB2xw) Tech stack: Python Why it matters: xmnlp 是一个功能强大的工具包，适用于各种文本处理任务，如情感分析、关键词提取、句子表征等，大大提高了文本处理的效率和准确性。

Overview: CodeFuse-ChatBot 是一个集成了多种技术的开源AI智能助手，旨在简化软件开发和运维过程，通过多智能体调度、代码整库分析、文档增强和垂类专属知识库等功能，提供智能化的开发运维解决方案。 Key features: - Multi-Agent Framework (codefuse-muAgent) - Code&Doc Repo RAG - Local Code Repository QA - Knowledge Base for DevOps Tech stack: Python Why it matters: 通过结合多智能体调度、代码整库分析、文档增强和垂类专属知识库等功能，CodeFuse-ChatBot 提供了一种全新的开发运维模式，旨在简化软件开发和运维流程，提高效率和质量。 Repository: codefuse-ai/codefuse-chatbot GitHub stars: 1295

Overview: This repository contains a skill designed to streamline iOS app development and testing, including Xcode builds and simulator interactions, with support for various AI agents. Key features: - Xcode builds via `xcodebuild` - Simulator interaction via `xcrun simctl` and `idb` - Optimized scripts for both human developers and AI agents Tech stack: Python 3.12+ Why it matters: This skill enhances the efficiency of iOS app development and testing, making it easier for developers and AI agents to build, test, and interact with apps, while also saving token waste with xcodebuild proxying. Repository: conorluddy/ios-simulator-skill GitHub stars: 1292

Overview: Difix3D+ is a state-of-the-art method for improving 3D reconstructions using single-step diffusion models, achieving significant performance gains over existing techniques. Key features: - Single-step diffusion models for 3D artifact removal - Improves 3D reconstructions with minimal post-processing - Achieves competitive results on challenging datasets Tech stack: Python Why it matters: Difix3D+ offers a novel approach to 3D reconstruction that significantly enhances the quality of rendered scenes, making it a valuable tool for researchers and practitioners in the field of computer vision and graphics.

### Overview: This repository is an awesome collection of face technology projects, focusing on various aspects such as face detection, alignment, recognition, and more. It serves as a comprehensive resource for developers and researchers interested in computer vision and facial recognition. ### Key features: - **Face Detection**: Tools and techniques for identifying faces in images. - **Face Alignment**: Algorithms for aligning faces to a standard template. - **Face Recognition**: Methods for comparing and identifying faces. - **Face Clustering**: Techniques for grouping similar faces. - **Face Expression**: Analysis and manipulation of facial expressions. - **Face Action**: Recognition and manipulation of facial actions. - **Face 3D**: Techniques for generating 3D models of faces. - **Face GAN**: Generative adversarial networks for face manipulation. - **Face Deblurring**: Methods for restoring blurred face images. - **Face Super-Resolution**: Enhancing the quality of face images. - **Face Manipulation**: Techniques for altering face images. - **Face Anti-Spoofing**: Methods for detecting fake or spoofed faces. - **Face Adversarial Attack**: Techniques for fooling face recognition systems. - **Face Cross-Modal**: Integration of face recognition with other modalities. - **Face Capture**: Tools for capturing and processing face data. ### Tech stack: The repository is built using HTML, CSS, and JavaScript. ### Why it matters: This repository is essential for developers and researchers looking to explore and implement cutting-edge face technology. It provides a wealth of resources and projects that can be used for

Overview: Endee.io is a high-performance vector database designed to handle up to 1B vectors on a single node, offering significant performance improvements through optimized indexing and execution. Key features: - Endee is available in the cloud. - Endee Enterprise is available under the Endee Enterprise license. - Free 30-day evaluation license available for testing. Tech stack: Endee is built on proprietary algorithms. Why it matters: Endee is optimized for large-scale vector data processing, making it suitable for applications requiring high performance and scalability.

### Overview: An awesome and curated list of the best code-LLMs for research. ### Key features: - 🚀 Top Code LLMs - 💡 Evaluation Toolkit - 🚀 Awesome Code LLMs Leaderboard - 📚 Awesome Code LLMs Papers - 🐢 Awesome Code Pre-Training Papers - 🐳 Awesome Code Instruction-Tuning Papers - 🐬 Awesome Code Alignment Papers - 🐟 Awesome Code Prompting Papers - 🐟 Awesome Code Benchmark \& Evaluation Papers ### Tech stack: Unknown ### Why it matters: An awesome and curated list of the best code-LLMs for research.

### Overview: StreamSpeech is an "All in One" seamless model designed for offline and simultaneous speech recognition, translation, and synthesis. It achieves state-of-the-art performance across various tasks. ### Key features: - **Offline**: Supports ASR, S2TT, S2ST, and TTS. - **Simultaneous**: Real-time ASR, S2TT, S2ST, and TTS. - **GUI Demo**: Offers a seamless model for ASR, translation, and synthesis. - **8 Tasks**: Supports offline and simultaneous speech recognition, translation, and synthesis. ### Tech stack: Python ### Why it matters: StreamSpeech provides a comprehensive solution for simultaneous communication, offering real-time translation and synthesis capabilities.

Overview: Bug0 AI is an open-source Playwright library designed for AI-driven browser regression testing, offering intelligent caching, auto-healing, and multi-model verification to help catch regressions early. Key features: - AI-driven browser regression testing - Intelligent caching - Auto-healing - Multi-model assertion verification Tech stack: TypeScript Why it matters: Bug0 AI streamlines the testing process by automating browser steps, leveraging AI models for natural language understanding, and ensuring tests remain stable without requiring frequent updates to prompts or retraining models. Repository: bug0inc/qa-agent GitHub stars: 1290 GitHub description: The open-source Playwright library for AI browser regression testing with intelligent caching, auto-healing, and multi-model verification.

Overview: BaseAI is a framework designed for building serverless and composable AI agents with memory and tools. It simplifies the process of creating AI pipelines and agents, allowing developers to focus on the core functionality without the complexity of frameworks. Key features: - Easy integration with any language - TypeScript-first framework - Supports memory and tools - Built-in agentic tools and memory (RAG) - Native support for Langbase API keys Tech stack: TypeScript Why it matters: BaseAI offers a more efficient and flexible approach to AI development compared to traditional frameworks, making it ideal for developers looking to build AI solutions quickly and without the constraints of framework limitations. Repository: CommandCodeAI/BaseAI GitHub stars: 1290 GitHub description: BaseAI — The Web AI Framework. The easiest way to build serverless autonomous AI agents with memory. The framework is moving towards the Langbase AI Primitives, emphasizing AI primitives over frameworks.