Best AI Tools

KittenTTS is an open-source, lightweight text-to-speech library that utilizes ONNX for efficient voice synthesis. The library offers multiple model sizes ranging from 15M to 80M parameters, allowing for high-quality audio output on standard CPUs. It includes built-in text preprocessing and support for eight distinct voices, making it suitable for various edge deployment scenarios.

LiteLLM is an open-source AI gateway that provides a unified interface for calling over 100 different LLM providers using the standard OpenAI format. It can be utilized as a Python SDK for direct integration or deployed as a proxy server to manage enterprise-grade features like load balancing and spend tracking. By abstracting provider-specific complexities, it allows developers to switch between models seamlessly without rewriting their existing code.

PyGWalker is a Python library that transforms pandas DataFrames into an interactive user interface for exploratory data analysis. It integrates the Graphic Walker visualization tool directly into Jupyter Notebooks, allowing users to create charts using simple drag-and-drop operations. The library supports various data cleaning, transformation, and visualization tasks, providing a seamless workflow for data scientists.

FlashMLA is a library of high-performance attention kernels developed by DeepSeek to power their V3 and V3.2-Exp models. It provides specialized implementations for both sparse and dense attention mechanisms during prefill and decoding stages. The library supports advanced features like FP8 KV caching and is optimized for modern GPU architectures to maximize computational throughput.

DeepEP is a specialized communication library designed to optimize Mixture-of-Experts and expert parallelism through high-throughput, low-latency GPU kernels. It supports both training and inference workloads by providing advanced features like asymmetric-domain bandwidth forwarding and hook-based communication-computation overlapping. The library is highly optimized for NVLink and RDMA environments, offering significant performance gains for large-scale distributed model operations.

DeepGEMM is a unified CUDA library providing high-performance tensor core kernels specifically optimized for modern large language models. It features a lightweight Just-In-Time compilation module that eliminates the need for complex CUDA installation builds. The library supports advanced operations including FP8/FP4 GEMMs, fused Mega MoE, and MQA scoring kernels for efficient GPU computation.

DeerFlow is an open-source super agent framework designed to orchestrate sub-agents, memory, and sandboxed environments for complex task execution. The platform features a ground-up rewrite in version 2.0, offering enhanced extensibility through a modular skill-based architecture. It supports diverse deployment options, including local development and Docker-based production environments, with integrated support for various LLM providers and messaging channels.

This project provides a systematic and beginner-friendly tutorial in Chinese, covering the official Anthropic programming tool Claude Code and the open-source AI assistant framework OpenClaw. The tutorial includes 25 in-depth guides, over 70 runnable code examples, and more than 170 FAQs, aiming to help developers quickly master AI programming and automated workflows. Content is updated in real-time to adapt to the latest versions, providing users with a learning path from zero-based knowledge to enterprise-level practice.

OpenSandbox is a versatile platform designed for AI applications that provides unified APIs and multi-language SDKs for managing sandbox environments. It supports both Docker and Kubernetes runtimes, enabling developers to execute code, manage files, and run commands in isolated containers. The platform also features robust security options, including support for gVisor and Firecracker, to ensure safe execution for various AI agent and training workloads.

DeerFlow is an open-source super agent framework designed to orchestrate sub-agents, memory, and sandboxed environments for complex task execution. The platform features a ground-up rewrite in version 2.0, offering enhanced extensibility through a modular skill-based architecture. It supports diverse deployment options, including local development and Docker-based production environments, with integrated support for various LLM providers and messaging channels.

PaddleOCR is a comprehensive toolkit designed to convert images and PDF documents into structured, LLM-ready data formats like Markdown and JSON. It features state-of-the-art vision-language models and high-performance text recognition engines that support over 100 languages. The platform is widely integrated into major AI agent and RAG frameworks, offering efficient deployment options across various hardware backends.

This project builds an executable thinking framework rather than a simple collection of quotes, based on Zhang Xuefeng's books, interviews, and decision records. It provides users with in-depth analysis for major selection, postgraduate entrance exam planning, and career development by distilling core mental models and decision heuristics. Users can integrate it into Claude Code to obtain practical advice from Zhang Xuefeng's perspective.

The Willow Inference Server allows users to self-host lightning-fast language inference tasks for various applications. It supports a range of functionalities including speech-to-text, text-to-speech, and large language model processing. Users can access official documentation and community discussions to optimize their experience with the platform.

Secret Llama is an entirely in-browser chatbot that allows users to run open-source models like Llama 3 and Mistral locally. Because the application operates directly within the browser, no conversation data ever leaves the user's computer. It provides a user-friendly interface similar to ChatGPT while functioning completely offline without requiring any server installations.

Tabby is an open-source, self-hosted AI coding assistant designed as a private alternative to GitHub Copilot. It operates entirely on-premises without requiring external cloud services or complex database management. The platform supports consumer-grade GPUs and integrates seamlessly with existing infrastructure via an OpenAPI interface.

Index-AniSora is a powerful open-source framework designed specifically for high-quality anime video generation and animation production. The system integrates a comprehensive data processing pipeline, a controllable generation model with spatiotemporal masking, and a specialized evaluation benchmark. It supports diverse creative tasks including image-to-video generation, character 3D modeling, style transfer, and multimodal guidance.

Xiaomi Miloco is an open-source smart home solution that leverages on-device large language models to integrate and control IoT devices. It utilizes visual data from home cameras to enable natural language interaction for complex task planning and scene understanding. The framework prioritizes privacy by processing video information locally while connecting seamlessly with the broader Xiaomi Home ecosystem.

ncnn is a high-performance neural network forward computation framework deeply optimized for mobile platforms. The framework has no third-party dependencies and features cross-platform capabilities, outperforming all known open-source frameworks on mobile CPUs. Developers can easily port deep learning models to mobile devices using ncnn to build various intelligent applications.

Slime is an LLM post-training framework designed to facilitate reinforcement learning scaling by integrating Megatron with SGLang. It provides high-performance training capabilities alongside a flexible data generation interface for custom workflows. The framework has been successfully utilized to power various research projects and production systems, including advanced reasoning and agentic models.

This tutorial provides users with a comprehensive guide to mastering OpenClaw from scratch, covering installation, deployment, core functional practice, and advanced automation techniques. The content is proofread based on the stable OpenClaw v2026.4.14 version and offers rich practical cases and troubleshooting tips. Through this tutorial, users can quickly build a personal AI work assistant to achieve a significant boost in personal efficiency.

AI Daily Digest is an automated tool that scrapes top technical blogs from Hacker News and utilizes AI for multi-dimensional scoring and summarization. It supports generating structured daily technical newsletters via command line or interactive chat, while providing trend summaries and visual statistics. The project is written in pure TypeScript with zero dependencies and supports flexible switching between Gemini, OpenAI, and other compatible API models.

This project successfully restores a readable source code directory containing 700,000 lines of code by reconstructing the source maps of Claude Code version 2.1.88. Developers can use this to conduct an in-depth study of the CLI tool's command system, the terminal UI built with React and Ink, and the implementation details of the MCP protocol. The project aims to provide architectural analysis and archival reference to help technical personnel understand its core business logic and engineering design.

OpenOcta is an open-source enterprise AI Agent runtime and control plane designed as a single Go binary for deployment anywhere. The project integrates core modules such as Gateway, Agent, Channels, and Cron, supporting seamless connection with business systems, APIs, and MCP tools. Developers can quickly perform task orchestration and automation integration via the built-in Control UI or CLI.

Tong Jincheng.skill is a Claude Code plugin built on approximately 200,000 words of primary video material, designed to analyze interpersonal relationships and emotional issues through the unique perspective of the 'Affectionate Grandmaster.' This tool is not a simple repeater of quotes, but rather extracts his street wisdom and cognitive framework to help users dissect life dilemmas in a direct and sincere manner. Users can activate the plugin with simple commands to gain decision-making insights regarding romance, personal growth, and social interaction.

This comprehensive guide provides a detailed walkthrough of the Hermes Agent framework developed by Nous Research. It covers core mechanisms like the self-improving learning loop, three-layer memory system, and automated skill evolution across seventeen chapters. The book serves as a practical resource for developers and AI enthusiasts looking to implement and customize autonomous AI agents.

Awesome-finance-skills is a plug-and-play collection of financial skills designed to provide large language models with real-time news, stock data, sentiment analysis, and market forecasting capabilities. The project supports various mainstream Agent frameworks, allowing users to quickly integrate financial analysis functions through simple command-line operations. Additionally, it provides logic chain visualization and professional research report generation tools to help AI agents achieve deep market insights.

Khazix Skills is an open-source collection of AI tools designed to transform the author's accumulated methodologies into reusable Prompts and Skills. The project provides lightweight prompt templates and structured instruction sets that follow the Agent Skills open standard. Users can integrate these tools into supported AI Agents via direct installation or manual configuration to enhance work efficiency.

This repository gathers 46 verified real-world application scenarios for OpenClaw AI agents, designed to help users improve work and life efficiency through automation. The content covers everything from beginner guides to deep adaptation use cases for domestic ecosystems such as Feishu, DingTalk, and WeCom. Each case provides detailed setup steps and copy-paste prompts, ensuring users can quickly get started and deploy personalized agents.

The OpenClaw Chinese Edition provides full Chinese interface support for the open-source personal AI assistant platform, covering both the CLI and the Dashboard web console. This project automatically synchronizes with official updates every hour, ensuring users can experience the latest features promptly. Users can quickly deploy via a simple installation wizard and utilize built-in ecosystem tools to achieve efficient AI interaction management.

ArcReel is an open-source video generation workbench powered by AI Agents, designed to automate the production of short videos from novel scripts. The platform orchestrates tasks through a multi-agent architecture and supports image and video generation capabilities from mainstream providers such as Gemini, Volcano Engine, Grok, and OpenAI. Users can manage projects via a visual interface and utilize asynchronous task queues and version rollback features to efficiently complete high-quality video creation.

Humanizer-zh is a skill tool designed for Claude Code that helps users identify and remove common AI-generated traces in text. The project analyzes 24 AI writing patterns to guide users in rewriting mechanical content into more natural and personalized human expressions. It not only provides automated rewriting features but also helps creators improve the authenticity and readability of their articles through specific writing principles.

FlashMLA is a library of high-performance attention kernels developed by DeepSeek to power their V3 and V3.2-Exp models. It provides specialized implementations for both sparse and dense attention mechanisms during prefill and decoding stages. The library supports advanced features like FP8 KV caching and is optimized for modern GPU architectures to maximize computational throughput.

ncnn is a high-performance neural network forward computation framework deeply optimized for mobile platforms. The framework has no third-party dependencies and features cross-platform capabilities, outperforming all known open-source frameworks on mobile CPUs. Developers can easily port deep learning models to mobile devices using ncnn to build various intelligent applications.

VoxCPM2 is a tokenizer-free, 2B parameter text-to-speech system that utilizes a diffusion autoregressive architecture to generate high-quality, expressive audio. The model supports 30 languages and offers advanced capabilities including voice design, controllable voice cloning, and studio-quality 48kHz output. It is fully open-source under the Apache-2.0 license and provides production-ready deployment options via vLLM-Omni and Nano-vLLM.

EvoCUA is a high-performance, open-source multimodal model designed for end-to-end computer automation tasks. It achieves top-tier results on the OSWorld benchmark and demonstrates strong cross-OS generalization capabilities on Windows. The model is recognized for its superior safety profile, maintaining the lowest unintended-behavior rate among leading computer-use agents.

Hermes Agent is a self-improving AI assistant designed by Nous Research that creates and refines skills through a built-in learning loop. It supports a wide range of LLM providers and can be deployed across various platforms including Telegram, Discord, and local terminal environments. The system features persistent memory, scheduled automations, and the ability to spawn subagents for complex, parallelized tasks.

Page Agent is a client-side tool that enables users to control web interfaces using natural language commands. It operates directly within the webpage using text-based DOM manipulation, eliminating the need for browser extensions, headless browsers, or multi-modal LLMs. The library supports flexible LLM integration and provides optional extensions for multi-page task automation.

WeKnora is an intelligent knowledge management and Q&A framework that utilizes LLMs to provide enterprise-grade document understanding and semantic retrieval. The platform offers both a RAG-based Quick Q&A mode for fast queries and a ReACT Agent engine for complex, multi-source reasoning tasks. It features a highly modular architecture that supports various document formats, multiple LLM providers, and seamless integration with popular IM channels for private or local deployment.

DeepEP is a specialized communication library designed to optimize Mixture-of-Experts and expert parallelism through high-throughput, low-latency GPU kernels. It supports both training and inference workloads by providing advanced features like asymmetric-domain bandwidth forwarding and hook-based communication-computation overlapping. The library is highly optimized for NVLink and RDMA environments, offering significant performance gains for large-scale distributed model operations.

OpenSandbox is a versatile platform designed for AI applications that provides unified APIs and multi-language SDKs for managing sandbox environments. It supports both Docker and Kubernetes runtimes, enabling developers to execute code, manage files, and run commands in isolated containers. The platform also features robust security options, including support for gVisor and Firecracker, to ensure safe execution for various AI agent and training workloads.

Khoj is a versatile personal AI application designed to extend user capabilities through advanced semantic search and document integration. It supports a wide range of local and online LLMs while offering flexible deployment options from on-device to cloud-scale enterprise solutions. Users can create custom agents and automate research tasks across various platforms including Obsidian, Emacs, and mobile devices.

code-review-graph builds a structural map of your codebase using Tree-sitter to provide AI assistants with precise, context-aware information. By tracking changes incrementally and calculating the blast radius of edits, it significantly reduces token consumption during code reviews. The tool integrates seamlessly with various AI coding platforms via the Model Context Protocol to ensure only relevant code is analyzed.

Kronos is an open-source decoder-only foundation model specifically designed to interpret and forecast financial K-line sequences. It utilizes a two-stage framework that quantizes multi-dimensional market data into hierarchical tokens before processing them through an autoregressive Transformer. The project provides pre-trained models, a specialized predictor class, and a complete pipeline for fine-tuning on custom financial datasets.

graphify is an AI coding assistant skill that builds a persistent knowledge graph from codebases, documents, images, and media files. It uses tree-sitter for deterministic code structure extraction and multimodal LLM agents to identify concepts and relationships across diverse file types. The resulting graph provides an interactive map and audit report that helps AI assistants navigate complex projects with significantly reduced token usage.

RTK is a high-performance CLI proxy that filters and compresses command outputs to significantly reduce LLM token consumption. It supports over 100 common commands and integrates seamlessly with popular AI coding tools via an auto-rewrite hook. By applying smart filtering, grouping, and truncation, it helps developers save up to 90% on token costs during AI-assisted development sessions.

The OpenAI Agents SDK is a lightweight framework designed for building complex multi-agent workflows. It supports a wide range of LLMs and provides essential features like tool integration, guardrails, and human-in-the-loop capabilities. Developers can also utilize sandbox agents for long-running tasks and leverage built-in tracing to debug and optimize their agentic applications.

Claude Code Local provides a suite of high-performance AI models that run entirely on Apple Silicon hardware without requiring cloud connectivity. The project features a native MLX server that enables local execution of Claude Code, browser automation, and voice interaction while ensuring complete data privacy. By eliminating outbound network calls and telemetry, it offers a secure, air-gapped environment for handling sensitive professional tasks.

Multica is an open-source platform designed to integrate coding agents into human teams as autonomous teammates. It provides a unified dashboard for task assignment, execution monitoring, and skill compounding across various agent runtimes. The system supports multi-workspace collaboration and allows teams to manage the full lifecycle of AI-driven development tasks.

World Monitor is an AI-powered platform that aggregates global news, geopolitical data, and infrastructure tracking into a unified situational awareness interface. It features a dual map engine and a comprehensive intelligence index to provide cross-stream correlation across various sectors. The project supports native desktop deployment via Tauri and offers multiple specialized variants for finance, technology, and commodity monitoring.

Voicebox is a comprehensive, local-first voice synthesis studio that allows users to clone voices and generate speech using seven different TTS engines. The platform supports 23 languages and includes a multi-track timeline editor for creating complex audio narratives with post-processing effects. Designed for privacy and performance, it runs natively on macOS, Windows, and Linux while providing a REST API for external integrations.

PPT Master is an open-source workflow that enables AI IDEs to generate fully editable PowerPoint presentations from various document formats. Unlike tools that export static images, this system creates files with real shapes, text boxes, and charts that users can modify directly. It prioritizes data privacy and flexibility by running locally and supporting multiple AI models without platform lock-in.

This project provides a set of structured guidelines designed to improve LLM coding behavior by addressing common pitfalls like overcomplication and making wrong assumptions. It implements four core principles—Think Before Coding, Simplicity First, Surgical Changes, and Goal-Driven Execution—to ensure more precise and reliable code generation. Users can easily integrate these rules into their development workflow via a Claude Code plugin, a CLAUDE.md file, or Cursor rules.

CC Switch is a desktop application that provides a centralized interface for managing multiple AI-powered CLI tools including Claude Code, Codex, and Gemini CLI. It eliminates the need for manual configuration file editing by offering over 50 provider presets and a visual dashboard for unified management. The tool also features cross-platform support, cloud synchronization, and built-in utilities for tracking usage and managing MCP servers.

CL4R1T4S provides a comprehensive repository of extracted system prompts, guidelines, and tools from major AI models and agents. The project aims to increase transparency by revealing the hidden instructions that shape how AI models behave and interact with users. It encourages community contributions to document and share these underlying prompt scaffolds for better understanding of AI intelligence.

Agent Skills provides a structured library of engineering workflows designed to help AI coding agents follow professional best practices throughout the development lifecycle. The project includes 20 specialized skills, ranging from spec-driven development to security hardening, which enforce consistency and quality in generated code. These skills are compatible with various AI agents and IDEs, ensuring that automated tasks meet production-grade standards.

This repository provides a curated collection of DESIGN.md files that allow AI agents to generate consistent, pixel-perfect user interfaces. Each document captures essential design elements like color palettes, typography, and component styling based on popular real-world websites. Developers can simply drop these files into their projects to guide AI coding agents in matching specific design aesthetics.

MemPalace provides a local-first solution for storing conversation history as verbatim text with structured semantic search capabilities. The system utilizes a pluggable backend architecture to organize data into wings, rooms, and drawers without requiring external API calls. It achieves high retrieval recall benchmarks while ensuring that all user data remains strictly on the local machine.

Taste Skill provides a collection of specialized instructions designed to improve the quality of frontend code generated by AI agents. The toolkit offers various design styles, ranging from minimalist to brutalist, which can be customized using adjustable parameters for variance and motion. It is framework-agnostic and integrates seamlessly with major AI coding tools to ensure premium, professional-grade interface outputs.

Caveman is a specialized plugin for AI agents that significantly reduces output token usage by enforcing a concise, telegraphic communication style. It maintains full technical accuracy while cutting approximately 75% of output tokens and 46% of input tokens through its compression tools. The project supports various agents including Claude Code, Cursor, and Gemini, offering multiple intensity levels and specialized modes like 文言文.

DeepTutor is an agent-native platform that provides a unified workspace for personalized learning through multiple interactive modes like chat, deep research, and quiz generation. It features autonomous TutorBots with persistent memory and a collaborative AI Co-Writer that integrates directly with your personal knowledge base. Users can deploy the system via a guided CLI, manual installation, or Docker to create a tailored educational environment.

Evolver is a GEP-powered engine designed to facilitate the self-evolution of AI agents by transforming ad hoc prompt adjustments into auditable, reusable assets. It functions by scanning runtime logs and memory to generate protocol-bound prompts that guide agent improvement without directly modifying source code. The system supports various operational modes, including standalone execution, continuous background loops, and integration with agent runtimes like OpenClaw and Cursor.

Claude Code Game Studios transforms a standard AI coding session into a structured game development environment using a hierarchical team of 49 specialized agents. The system provides 72 slash commands and automated validation hooks to enforce professional design, coding, and production standards throughout the development lifecycle. Users maintain full control over the project while benefiting from AI-driven organization, quality gates, and domain-specific expertise.

Omi is an open-source platform that functions as a digital second brain by capturing and transcribing your screen and conversations in real-time. It provides AI-driven summaries, action items, and a conversational interface that remembers everything you have seen or heard. The system supports cross-platform integration across desktop, mobile, and specialized wearable hardware.

JaQMC is a modular framework built on JAX that utilizes deep neural networks as variational wavefunctions to solve the electronic Schrödinger equation. It provides a flexible environment for modeling interacting electrons in atoms, molecules, and solid-state systems without relying on traditional basis sets. The platform supports advanced architectures like FermiNet and PsiFormer while leveraging JIT compilation and multi-device parallelism for high-performance simulations.

vLLM Kunlun is a community-maintained plugin that enables seamless execution of vLLM on Kunlun XPU hardware. It utilizes a hardware-pluggable interface to decouple the Kunlun backend from the core vLLM framework. This integration allows users to run a wide range of Transformer, Mixture-of-Expert, and multimodal models efficiently on Kunlun3 P800 hardware.

ERNIE-Image is an open-source text-to-image model developed by Baidu based on the single-stream Diffusion Transformer (DiT) architecture. The model is equipped with a lightweight prompt enhancer that can expand short user inputs into structurally rich and detailed descriptions. With an 8B parameter scale, it demonstrates exceptional performance in text rendering and instruction following while maintaining efficient deployment.

RTP-LLM is a high-performance large language model inference acceleration engine developed by the Alibaba Foundation Model Inference Team. This engine has been widely applied in various Alibaba business scenarios such as Taobao and Tmall, and it supports multiple mainstream model formats and hardware architectures. By integrating advanced operator optimization, quantization techniques, and distributed inference capabilities, it provides developers with efficient and flexible deployment solutions.

ANOLISA is an evolution of Anolis OS designed specifically to serve as an operating system for AI agent workloads. It provides a comprehensive architecture including a copilot shell, a security kernel, and observability tools based on eBPF. The platform offers a curated library of system administration and DevOps skills to facilitate efficient AI agent management.

TorchEasyRec is a PyTorch-based framework designed for building production-ready deep learning recommendation models. It supports a wide range of tasks including candidate generation, ranking, multi-task learning, and generative recommendation. The framework provides flexible configuration, distributed training capabilities, and seamless integration with various data sources and deployment environments.

PaddlePaddle is a comprehensive industrial deep learning platform that provides core frameworks, model libraries, and end-to-end development tools. It supports advanced features like unified dynamic and static graphs, automatic parallelism, and high-order differentiation for scientific computing. The platform is widely adopted across various industrial sectors to help developers and companies efficiently commercialize AI solutions.

Claude-Mem is a persistent memory compression system designed to maintain context across sessions for Claude Code and similar CLI tools. It automatically captures tool usage and generates semantic summaries to ensure continuity of project knowledge. The system includes a web viewer, hybrid search capabilities, and fine-grained privacy controls for developers.

This repository provides a comprehensive collection of best practices, implementation patterns, and workflows for optimizing Claude Code. It covers essential concepts like subagents, commands, skills, and orchestration to help developers transition from basic coding to advanced agentic engineering. Users can explore various development workflows and integration strategies to enhance their productivity and automation capabilities.

CLIProxyAPI is a versatile proxy server that provides OpenAI, Gemini, and Claude-compatible API interfaces for various command-line tools. It supports OAuth-based authentication for models like OpenAI Codex and Claude Code, enabling seamless integration with local or multi-account setups. The project also features advanced load balancing, model mapping, and a reusable Go SDK for embedding proxy capabilities into custom applications.

T3 Code provides a minimal web-based graphical user interface designed specifically for interacting with coding agents. The platform currently supports integration with Codex and Claude, with plans to expand support for additional providers in the future. Users can access the tool via a desktop application or run it directly using npx for quick deployment.

Magika is an AI-powered tool that utilizes deep learning to provide highly accurate file type identification for over 200 content types. It features a highly optimized model that delivers inference results in milliseconds while maintaining approximately 99% accuracy. The project offers a versatile command-line interface and language bindings for Python, JavaScript, and Rust to support diverse developer workflows.

pi-autoresearch is an extension for the pi AI coding agent that enables autonomous optimization loops by testing, benchmarking, and refining code changes. It supports various optimization targets such as test speed, bundle size, and LLM training metrics through a persistent session workflow. The tool includes a live dashboard, confidence scoring to filter out noise, and the ability to finalize experiments into clean, reviewable branches.

Chrome DevTools for Agents is an MCP server that enables AI coding assistants to control and inspect live Chrome browser instances. It provides a comprehensive suite of tools for browser automation, performance analysis, and in-depth debugging. The project supports seamless integration with various AI coding platforms to enhance developer workflows through reliable browser interaction.

FastDeploy is a professional large language model and vision-language model inference deployment toolkit based on PaddlePaddle, designed to provide out-of-the-box production-grade deployment solutions. The toolkit supports various mainstream hardware platforms and integrates advanced acceleration technologies such as load balancing, unified KV cache transmission, and full quantization format support. Developers can achieve rapid deployment through OpenAI API-compatible interfaces, thereby significantly improving model inference throughput and resource utilization.

Open Agents is an open-source reference application designed for building and running background coding agents on the Vercel platform. The system utilizes a three-layer architecture that separates the web interface, durable agent workflows, and isolated sandbox execution environments. This modular design allows developers to perform complex coding tasks, such as repository management and automated pull requests, without requiring active local machine involvement.

This repository provides a comprehensive educational guide on building effective harnesses for AI agents by separating model intelligence from environmental infrastructure. It features twelve progressive sessions that teach developers how to implement tools, context management, and task coordination to support autonomous agent behavior. By reverse-engineering the architecture of Claude Code, the project demonstrates how to create robust systems that allow models to perceive, reason, and act within specific domains.

This comprehensive course provides over 260 lessons across 20 phases to help developers master AI engineering from foundational mathematics to advanced autonomous agent swarms. It utilizes an AI-native learning approach where students work alongside coding agents to build, test, and deploy reusable tools. The curriculum spans multiple programming languages including Python, TypeScript, Rust, and Julia to ensure practical, production-ready skills.

Impeccable is a comprehensive design skill and command set designed to guide AI agents toward superior frontend UI development. It provides domain-specific references and curated anti-patterns to help developers avoid common design mistakes like generic templates and poor accessibility. The tool supports various AI coding environments and includes a standalone CLI for detecting design issues across projects.

Claude How To provides a structured, visual learning path to help developers master the full capabilities of Claude Code. The guide offers copy-paste templates, interactive quizzes, and detailed tutorials covering everything from slash commands to complex agent workflows. It is designed to bridge the gap between basic feature knowledge and the ability to build production-ready automation pipelines.

GBrain is an autonomous memory system designed to provide AI agents with a persistent, evolving knowledge base that compounds in intelligence over time. It utilizes a "thin harness, fat skills" architecture to ingest, enrich, and organize data from various sources like meetings, emails, and notes. The system features a hybrid search engine and 25 specialized skills that allow agents to perform complex tasks, maintain citations, and manage information autonomously.

The Pi Monorepo provides a comprehensive suite of tools designed for building AI agents and managing LLM deployments. It includes various packages for agent runtimes, multi-provider LLM APIs, and interactive coding agent interfaces. Developers can also contribute to the ecosystem by sharing open-source coding agent sessions to help improve real-world model performance.

ROCK is a scalable environment management framework designed specifically for agentic reinforcement learning applications. It utilizes a client-server architecture to provide secure, isolated sandbox environments for development, testing, and research. The platform supports standardized interfaces like the GEM protocol and offers a unified SDK for seamless interaction with reinforcement learning agents.

PaddleX is a low-code development tool built on the PaddlePaddle framework, integrating over 200 pre-trained models and 33 model pipelines. It supports the entire development process from model training to inference and is compatible with various mainstream domestic and international hardware. Developers can quickly implement and deploy industrial-grade AI applications using a minimalist Python API or a graphical interface.

Weft is a programming language designed to integrate LLMs, human interactions, and infrastructure into a unified, visual workflow. It features durable execution to ensure programs survive crashes and supports complex logic through a typed, modular node system. Developers can build and manage sophisticated agentic systems by wiring together native nodes without the need for manual plumbing.

fireworks-tech-graph enables users to generate professional SVG and PNG technical diagrams directly from natural language descriptions. The tool supports 14 UML diagram types and includes 7 distinct visual styles tailored for various documentation needs. It is specifically optimized for AI and agent-based domain patterns, allowing for rapid visualization without manual drawing.

OpenDataLoader PDF is a high-performance, open-source parser designed to convert PDF documents into structured formats like Markdown, JSON, and HTML for AI and RAG pipelines. It features a hybrid processing mode that combines deterministic local parsing with AI-driven analysis to achieve industry-leading extraction accuracy for complex tables, formulas, and scanned documents. Additionally, the project provides automated accessibility solutions, including end-to-end Tagged PDF generation compliant with international standards.

OpenMontage is an open-source, agentic system that transforms AI coding assistants into comprehensive video production studios. It automates the entire creative workflow, including research, scripting, asset generation, editing, and final composition. The platform supports both AI-generated visuals and real-footage documentary montages using a variety of free and premium tools.

oh-my-codex is a workflow enhancement layer designed to improve the daily runtime experience of the OpenAI Codex CLI. It provides structured processes for task clarification, planning, and execution through specialized commands like deep-interview and team-based parallel processing. The tool maintains project state, logs, and memory in a local .omx directory to ensure consistent and durable development workflows.

AgentKit Code Workshop is a companion sample repository for the AI Agent development platform launched by Volcano Engine, designed to help developers quickly master the agent construction and deployment process. The project provides a variety of code examples ranging from basic introductions to complex business scenarios, covering core functions such as multi-agent collaboration, RAG retrieval augmentation, and tool calling. Developers can use these tutorials to gain an in-depth understanding of the AgentKit development toolchain, thereby efficiently implementing various intelligent applications.

PaddleFormers is a Transformers library built on the PaddlePaddle framework, designed to provide training interfaces for Large Language Models and Vision-Language Models equivalent to Hugging Face. By integrating tensor parallelism, pipeline parallelism, and automatic mixed precision, the project achieves training performance that surpasses Megatron-LM on key models. Furthermore, it fully supports the Safetensors format and is deeply adapted to various domestic computing chips, helping developers efficiently complete the full model training process.

colleague.skill allows users to create AI personas based on the work habits and personalities of their colleagues by processing various communication data sources. The system utilizes a two-part architecture that combines professional work capabilities with nuanced behavioral traits to simulate realistic interactions. Users can easily manage, update, and roll back these AI skills to maintain accurate representations of their team members' expertise.

Camofox-browser is a specialized server designed to provide AI agents with reliable web browsing capabilities by leveraging the Camoufox engine for C++ level fingerprint spoofing. It offers a REST API that simplifies interactions through accessibility snapshots, stable element references, and built-in search macros. The system is optimized for efficiency and deployment, featuring automatic idle shutdown and session isolation to support scalable agent operations.

get-shit-done is a spec-driven development system designed to maintain high code quality by preventing context rot in AI coding assistants. It orchestrates subagents to handle project planning, research, and execution while maintaining clean git history and atomic commits. The system provides a structured workflow for developers to build complex features consistently without the overhead of enterprise project management.

TimesFM is a decoder-only foundation model developed by Google Research specifically for time-series forecasting tasks. The latest 2.5 version features a 200M parameter architecture that supports up to 16k context length and continuous quantile forecasting. The repository provides comprehensive tools for inference, fine-tuning with LoRA, and integration with agentic workflows.

Paseo provides a unified interface to manage and run various coding agents like Claude Code, Codex, and OpenCode on your local machine. It supports cross-device workflows, allowing users to interact with agents through desktop, mobile, web, or CLI applications. The platform prioritizes privacy by operating without telemetry or forced logins while enabling powerful agent orchestration capabilities.

oh-my-claudecode provides a multi-agent orchestration layer designed to enhance the Claude Code experience with zero learning curve. It enables advanced features like team-based task execution, intelligent model routing, and persistent autonomous workflows directly within your terminal. The tool simplifies complex development tasks by automating delegation, parallelization, and Socratic requirement clarification.

Tair KVCache is an Alibaba Cloud system designed to accelerate Large Language Model inference through distributed memory pooling and dynamic multi-level caching. The project provides a centralized manager for unified metadata handling and a simulation tool for predicting performance metrics without requiring GPU resources. These components work together to improve inference efficiency while reducing overall infrastructure costs.

This tutorial provides users with a comprehensive guide to mastering OpenClaw from scratch, covering installation, deployment, core functional practice, and advanced automation techniques. The content is proofread based on the stable OpenClaw v2026.4.14 version and offers rich practical cases and troubleshooting tips. Through this tutorial, users can quickly build a personal AI work assistant to achieve a significant boost in personal efficiency.

This project successfully restores a readable source code directory containing 700,000 lines of code by reconstructing the source maps of Claude Code version 2.1.88. Developers can use this to conduct an in-depth study of the CLI tool's command system, the terminal UI built with React and Ink, and the implementation details of the MCP protocol. The project aims to provide architectural analysis and archival reference to help technical personnel understand its core business logic and engineering design.

Humanizer-zh is a skill tool designed for Claude Code that helps users identify and remove common AI-generated traces in text. The project analyzes 24 AI writing patterns to guide users in rewriting mechanical content into more natural and personalized human expressions. It not only provides automated rewriting features but also helps creators improve the authenticity and readability of their articles through specific writing principles.

PaddlePaddle is a comprehensive industrial deep learning platform that provides core frameworks, model libraries, and end-to-end development tools. It supports advanced features like unified dynamic and static graphs, automatic parallelism, and high-order differentiation for scientific computing. The platform is widely adopted across various industrial sectors to help developers and companies efficiently commercialize AI solutions.

This project provides a systematic and beginner-friendly tutorial in Chinese, covering the official Anthropic programming tool Claude Code and the open-source AI assistant framework OpenClaw. The tutorial includes 25 in-depth guides, over 70 runnable code examples, and more than 170 FAQs, aiming to help developers quickly master AI programming and automated workflows. Content is updated in real-time to adapt to the latest versions, providing users with a learning path from zero-based knowledge to enterprise-level practice.

This repository gathers 46 verified real-world application scenarios for OpenClaw AI agents, designed to help users improve work and life efficiency through automation. The content covers everything from beginner guides to deep adaptation use cases for domestic ecosystems such as Feishu, DingTalk, and WeCom. Each case provides detailed setup steps and copy-paste prompts, ensuring users can quickly get started and deploy personalized agents.

The OpenClaw Chinese Edition provides full Chinese interface support for the open-source personal AI assistant platform, covering both the CLI and the Dashboard web console. This project automatically synchronizes with official updates every hour, ensuring users can experience the latest features promptly. Users can quickly deploy via a simple installation wizard and utilize built-in ecosystem tools to achieve efficient AI interaction management.

MedgeClaw is an open-source biomedical research assistant that integrates OpenClaw and Claude Code to automate complex scientific workflows. Users can interact with the system via messaging platforms like WhatsApp, Slack, or Discord to trigger analyses in R and Python environments. The platform provides a real-time research dashboard for monitoring progress, viewing code, and accessing interactive outputs.

AI Daily Digest is an automated tool that scrapes top technical blogs from Hacker News and utilizes AI for multi-dimensional scoring and summarization. It supports generating structured daily technical newsletters via command line or interactive chat, while providing trend summaries and visual statistics. The project is written in pure TypeScript with zero dependencies and supports flexible switching between Gemini, OpenAI, and other compatible API models.

Slime is an LLM post-training framework designed to facilitate reinforcement learning scaling by integrating Megatron with SGLang. It provides high-performance training capabilities alongside a flexible data generation interface for custom workflows. The framework has been successfully utilized to power various research projects and production systems, including advanced reasoning and agentic models.

Chinese-novelist is a skill plugin designed for Claude Code, aimed at helping users complete the entire process of writing Chinese novels through simple interactions. Users only need to answer five core questions, and the AI can automatically generate detailed outlines, character profiles, and coherent chapter content. The tool incorporates professional writing principles and quality checklists to ensure the coherence and appeal of the novel's plot.

Chinese-novelist is a skill plugin designed for Claude Code, aimed at helping users complete the entire process of writing Chinese novels through simple interactions. Users only need to answer five core questions, and the AI can automatically generate detailed outlines, character profiles, and coherent chapter content. The tool incorporates professional writing principles and quality checklists to ensure the coherence and appeal of the novel's plot.

MedgeClaw is an open-source biomedical research assistant that integrates OpenClaw and Claude Code to automate complex scientific workflows. Users can interact with the system via messaging platforms like WhatsApp, Slack, or Discord to trigger analyses in R and Python environments. The platform provides a real-time research dashboard for monitoring progress, viewing code, and accessing interactive outputs.

Thunderbolt is an open-source, cross-platform AI client designed for on-premise deployment and data ownership. It supports a wide range of frontier, local, and on-premise models across desktop and mobile environments. The project is currently under active development with a focus on enterprise readiness and security.

OpenCode is a free, open-source AI practical course designed for beginners, aiming to help users master methods to improve work efficiency using AI within 4 hours. The tutorial provides in-depth Chinese content and supports direct connection to mainstream domestic models without complex network configurations. The course covers five stages from quick start to deep customization, and provides rich practical projects and Prompt templates for learners to use.

ROLL is an efficient, user-friendly reinforcement learning library specifically designed for training and scaling Large Language Models on large-scale GPU clusters. It utilizes a multi-role distributed architecture powered by Ray to support complex tasks like human preference alignment, reasoning, and agentic interactions. The framework integrates advanced technologies such as Megatron-Core, vLLM, and SGLang to accelerate model training and inference across diverse hardware environments.

Awesome-finance-skills is a plug-and-play collection of financial skills designed to provide large language models with real-time news, stock data, sentiment analysis, and market forecasting capabilities. The project supports various mainstream Agent frameworks, allowing users to quickly integrate financial analysis functions through simple command-line operations. Additionally, it provides logic chain visualization and professional research report generation tools to help AI agents achieve deep market insights.

FastDeploy is a professional large language model and vision-language model inference deployment toolkit based on PaddlePaddle, designed to provide out-of-the-box production-grade deployment solutions. The toolkit supports various mainstream hardware platforms and integrates advanced acceleration technologies such as load balancing, unified KV cache transmission, and full quantization format support. Developers can achieve rapid deployment through OpenAI API-compatible interfaces, thereby significantly improving model inference throughput and resource utilization.

The Claude Cookbooks provide a comprehensive collection of code snippets and guides to help developers integrate Claude into their own applications. The repository covers a wide range of topics including tool use, multimodal capabilities, and advanced techniques like prompt caching. These resources are designed to be easily adaptable for various programming languages and project requirements.

Protenix is an open-source framework designed for high-accuracy biomolecular structure prediction, offering models that perform competitively with state-of-the-art methods. The project provides multiple versions, including the enhanced Protenix-v2, which demonstrates significant improvements in antibody-antigen structure prediction and ligand-related plausibility. It is released under the Apache 2.0 license, making it freely accessible for both academic and commercial research applications.

AutoFlow is an open-source knowledge base tool that utilizes graph RAG technology built on TiDB Vector, LlamaIndex, and DSPy. The platform provides a Perplexity-style conversational search experience powered by an advanced built-in website crawler. Users can also integrate a customizable search widget into their own websites using a simple JavaScript snippet.

Index-AniSora is a powerful open-source framework designed specifically for high-quality anime video generation and animation production. The system integrates a comprehensive data processing pipeline, a controllable generation model with spatiotemporal masking, and a specialized evaluation benchmark. It supports diverse creative tasks including image-to-video generation, character 3D modeling, style transfer, and multimodal guidance.

Xiaomi Miloco is an open-source smart home solution that leverages on-device large language models to integrate and control IoT devices. It utilizes visual data from home cameras to enable natural language interaction for complex task planning and scene understanding. The framework prioritizes privacy by processing video information locally while connecting seamlessly with the broader Xiaomi Home ecosystem.

RTP-LLM is a high-performance large language model inference acceleration engine developed by the Alibaba Foundation Model Inference Team. This engine has been widely applied in various Alibaba business scenarios such as Taobao and Tmall, and it supports multiple mainstream model formats and hardware architectures. By integrating advanced operator optimization, quantization techniques, and distributed inference capabilities, it provides developers with efficient and flexible deployment solutions.

Pipcook is a modular JavaScript application framework designed to help front-end engineers integrate machine learning into their workflows. It provides a comprehensive pipeline system that allows users to train, validate, and deploy machine learning models directly within the Node.js environment. By bridging access to Python packages, the framework enables developers to leverage powerful machine learning tools without requiring deep expertise in the field.

Pipcook is a modular JavaScript application framework designed to help front-end engineers integrate machine learning into their workflows. It provides a comprehensive pipeline system that allows users to train, validate, and deploy machine learning models directly within the Node.js environment. By bridging access to Python packages, the framework enables developers to leverage powerful machine learning tools without requiring deep expertise in the field.

The PaddlePaddle community serves as a central hub for developers to contribute to the framework through code improvements, documentation, and presentations. It provides structured governance, specialized working groups, and various mentorship programs to support active participation. Contributors are recognized through official certifications, release notes, and inclusion in the project's authorship records.

EvoCUA is a high-performance, open-source multimodal model designed for end-to-end computer automation tasks. It achieves top-tier results on the OSWorld benchmark and demonstrates strong cross-OS generalization capabilities on Windows. The model is recognized for its superior safety profile, maintaining the lowest unintended-behavior rate among leading computer-use agents.

vLLM Kunlun is a community-maintained plugin that enables seamless execution of vLLM on Kunlun XPU hardware. It utilizes a hardware-pluggable interface to decouple the Kunlun backend from the core vLLM framework. This integration allows users to run a wide range of Transformer, Mixture-of-Expert, and multimodal models efficiently on Kunlun3 P800 hardware.

This tool is officially maintained by Paddle and aims to achieve efficient automated migration from PyTorch code to PaddlePaddle code. It supports one-click conversion of over 1,600 PyTorch APIs and 200 torchvision APIs, maintaining an average conversion rate of over 95% in tests. The conversion process is operated via the command line, preserves the style and structure of the original code, and provides detailed conversion logs and summaries.

ROCK is a scalable environment management framework designed specifically for agentic reinforcement learning applications. It utilizes a client-server architecture to provide secure, isolated sandbox environments for development, testing, and research. The platform supports standardized interfaces like the GEM protocol and offers a unified SDK for seamless interaction with reinforcement learning agents.

ANOLISA is an evolution of Anolis OS designed specifically to serve as an operating system for AI agent workloads. It provides a comprehensive architecture including a copilot shell, a security kernel, and observability tools based on eBPF. The platform offers a curated library of system administration and DevOps skills to facilitate efficient AI agent management.

The AI Website Cloner Template is a reusable framework designed to reverse-engineer existing websites into modern Next.js codebases using AI coding agents. By providing a target URL, the system performs a multi-phase pipeline that extracts design tokens, assets, and component specifications to reconstruct the site. It supports a wide range of AI agents and utilizes a tech stack including Next.js 16, Tailwind CSS v4, and shadcn/ui.

PaddleCustomDevice is a custom hardware integration solution provided by the PaddlePaddle framework. This project aims to help developers integrate various third-party hardware backends into the PaddlePaddle ecosystem. It currently supports multiple mainstream hardware backends, including Ascend, Cambricon, Intel GPU, and Apple MPS.

Google AI Edge Gallery is a mobile application designed to run powerful open-source Large Language Models directly on your device. It offers a fully offline and private environment for users to experience advanced generative AI capabilities, including the latest Gemma 4 family. The app provides a comprehensive suite of tools for model management, benchmarking, and interactive AI features.

PaddleCustomDevice is a custom hardware integration solution provided by the PaddlePaddle framework. This project aims to help developers integrate various third-party hardware backends into the PaddlePaddle ecosystem. It currently supports multiple mainstream hardware backends, including Ascend, Cambricon, Intel GPU, and Apple MPS.

Memvid is a database-free, single-file memory layer designed to provide AI agents with instant retrieval and long-term memory capabilities. Through an innovative "smart frame" design, it encapsulates data, embeddings, and indexes into a single file, achieving efficient compression and parallel reading. The system is model-agnostic and requires zero infrastructure dependencies, supporting persistent memory in various offline or online scenarios.

Ralph is an autonomous AI agent loop that repeatedly executes coding tasks using tools like Amp or Claude Code until all project requirements are met. Each iteration operates within a fresh context, maintaining project state through git history, progress logs, and a structured JSON task list. The system ensures continuous progress by breaking down large features into manageable user stories that are verified through automated quality checks.

CorridorKey is a neural network-based tool designed to solve the complex problem of unmixing foreground subjects from green screen backgrounds. By predicting the true straight color and a clean linear alpha channel for every pixel, it preserves delicate details like motion blur and transparency that traditional keyers often destroy. The software supports high-fidelity VFX workflows by outputting 16-bit and 32-bit linear float EXR files compatible with industry-standard compositing applications.

The last30days tool is an AI agent-led search engine that synthesizes real-time data from social media, developer platforms, and prediction markets to provide current insights. By bridging disconnected platforms like Reddit, X, GitHub, and YouTube, the agent scores information based on actual human engagement rather than traditional SEO metrics. It functions as an expert research assistant that delivers concise, evidence-based briefs on any topic, person, or company from the past month.

Obsidian-skills provides a collection of tools designed to integrate Obsidian functionality with compatible AI agents. These skills adhere to the Agent Skills specification, allowing for seamless use with platforms like Claude Code, Codex CLI, and OpenCode. Users can install the plugin through various methods including marketplace commands, npx, or manual directory configuration.

This comprehensive guide provides a detailed walkthrough of the Hermes Agent framework developed by Nous Research. It covers core mechanisms like the self-improving learning loop, three-layer memory system, and automated skill evolution across seventeen chapters. The book serves as a practical resource for developers and AI enthusiasts looking to implement and customize autonomous AI agents.

PersonaPlex is a real-time, full-duplex speech-to-speech model built on the Moshi architecture that enables precise persona control through text prompts and audio voice conditioning. The model is trained on a mix of synthetic and real-world conversational data to deliver natural, low-latency interactions. Users can deploy the model via a provided server interface or perform offline evaluations using specific voice embeddings and role-based prompts.

Waza provides a collection of Claude Code skills designed to translate essential engineering habits into executable AI workflows. By focusing on specific, high-impact techniques rather than bloated configurations, it helps developers maintain high standards for design, debugging, and documentation. These skills are built from real-world project data to ensure Claude operates with the precision and intentionality of an experienced engineer.

VibeVoice is a collection of open-source voice AI models that utilize continuous speech tokenizers and a next-token diffusion framework to achieve high-fidelity audio processing. The project provides specialized models for long-form automatic speech recognition, real-time streaming text-to-speech, and multi-speaker synthesis. These models are designed for research purposes, offering capabilities like single-pass processing for hour-long audio and support for over 50 languages.

MiroFish is a next-generation AI prediction engine based on multi-agent technology that constructs high-fidelity digital parallel worlds by extracting real-world seed information. Users can perform simulations within this sandbox by injecting variables, thereby precisely deducing future trajectories. The platform aims to provide decision-makers with a zero-risk testing laboratory while offering individual users a creative simulation space.

Obsidian Mind is an Obsidian knowledge base template designed specifically for Claude Code, aiming to solve the problem of AI long-term memory loss through automated conversation hooks and structured storage. It ensures that Claude has access to the complete context at the start of every session by automatically linking work notes, decision records, meeting minutes, and performance evidence into a knowledge graph. The system supports daily workflow management through natural language interaction and can automatically generate performance evaluation briefs and project summaries.

agent-browser is a high-performance browser automation command-line tool built with Rust, specifically designed for AI agents. It supports web interaction, element localization, and state management through simple commands, eliminating the need for complex Playwright or Node.js environments. The tool provides extensive session persistence, authentication management, and debugging features to ensure that AI agents can operate safely and efficiently.

Rowboat is an open-source AI coworker that integrates with your email and meeting notes to build a persistent, local knowledge graph. It utilizes this context to assist with tasks like drafting documents, preparing for meetings, and tracking projects while keeping all data in an editable Markdown format. By maintaining long-lived memory on your machine, it allows for compounding context that improves over time without relying on external cloud storage.

llmfit is a terminal-based utility that analyzes your system's hardware to identify which large language models will run effectively on your specific configuration. It provides an interactive TUI and CLI to score models based on quality, speed, and memory fit while supporting various backends like Ollama, llama.cpp, and MLX. Users can also perform hardware simulations to test how different model configurations would perform on target system specifications.

Claudian is an Obsidian plugin that integrates AI coding agents like Claude Code and Codex directly into your vault. It transforms your vault into an active working directory where agents can read, write, search, and execute bash commands. Users can interact with these agents through a chat sidebar, inline editing, and support for Model Context Protocol servers.

Feynman is an open-source AI research agent designed to assist with complex tasks like literature reviews, paper auditing, and experiment replication. The platform utilizes a multi-agent system to gather evidence, perform simulated peer reviews, and draft structured research findings. Users can interact with the tool via a terminal interface, supporting both local and cloud-based execution environments.

Ghost Pepper is a privacy-focused macOS application that provides local speech-to-text transcription without relying on cloud APIs. Users simply hold the Control key to record audio, which is then transcribed and automatically pasted into any text field. The app utilizes local models for both speech recognition and text cleanup to ensure that no data ever leaves the user's machine.

LiteRT-LM is a high-performance, production-ready inference framework designed by Google for deploying Large Language Models on edge devices. It supports a wide range of platforms including Android, iOS, desktop, and IoT, while leveraging GPU and NPU hardware acceleration for optimal performance. The framework enables advanced capabilities such as multi-modality and function calling, powering on-device AI experiences in various Google products.

Oh My OpenCode is an open-source agent tool designed to break free from single-model lock-in and achieve efficient development by orchestrating multiple AI models. It enables parallel task processing and automated execution through the introduction of Discipline Agents and the ultrawork command, eliminating the need for manual model switching. The tool significantly improves code modification accuracy and the development experience through its Hash-Anchored Edit Tool and deep initialization features.

This project contains the complete leaked source code for Anthropic's official Claude Code CLI tool, which was discovered on March 31, 2026, via map files in the npm package. The repository provides the original source code built with TypeScript and Bun, accompanied by detailed architectural documentation and exploration guides. Users can interactively delve into the tool's internal implementation and design patterns through the built-in MCP server.

This plugin allows Claude Code users to invoke Codex directly within their workflow for code reviews or task delegation. Users can execute read-only reviews, adversarial reviews, and background task management through a series of slash commands. The tool leverages the locally installed Codex CLI and existing authentication configurations, ensuring seamless integration with the user's current development environment.

Qwen Code is an open-source AI agent designed to operate directly within the terminal to help developers understand codebases and automate tasks. It supports multiple authentication methods, including Qwen OAuth and various OpenAI-compatible API providers, to offer flexible model integration. The tool provides a feature-rich agentic workflow and can be integrated into popular IDEs like VS Code, Zed, and JetBrains.

Clicky is an open-source AI teaching assistant that integrates directly into your macOS environment to provide real-time guidance. The application uses screen recording, voice interaction, and cursor control to act as a virtual tutor that can see and interact with your desktop. Users can deploy the project locally by configuring a Cloudflare Worker proxy and building the Swift-based application via Xcode.

QMD is an on-device search engine that indexes markdown notes, documentation, and transcripts for efficient local retrieval. It utilizes a hybrid approach combining BM25 full-text search, vector semantic search, and LLM-based re-ranking to deliver high-quality results. The tool is designed for agentic workflows, offering both a command-line interface and an MCP server for seamless integration with AI agents.

OpenSpec is a lightweight standardization framework designed to improve the predictability and organization of AI-assisted development by establishing clear specifications before coding. It introduces an artifact-guided workflow that allows developers and AI to reach a consensus on requirements, design, and tasks before writing code. The tool supports over 20 AI coding assistants and can flexibly adapt to various development scenarios ranging from personal projects to enterprise-level applications.

The AI Hedge Fund is an educational proof-of-concept project designed to explore how artificial intelligence can be utilized for making trading decisions. It employs a multi-agent system that simulates various famous investment strategies and analytical approaches to evaluate stocks. The system is strictly for research purposes and does not execute real-world financial trades.

Archon is an open-source workflow engine that allows developers to define AI coding processes using deterministic YAML workflows. By structuring tasks like planning, implementation, and validation, it ensures that AI-driven development is repeatable, isolated, and reliable across projects. Users can compose workflows that mix deterministic operations with AI-driven steps to automate complex software development tasks.

JoyAI-Image is a unified multimodal foundation model that integrates an 8B Multimodal Large Language Model with a 16B Multimodal Diffusion Transformer to support image understanding, generation, and editing. The model utilizes a closed-loop collaboration between understanding and generation to enhance spatial reasoning and controllable editing capabilities. It provides a scalable training pipeline and supports advanced features like multi-view generation and precise spatial manipulation.

GitNexus indexes codebases into a comprehensive knowledge graph to provide AI agents with deep architectural context, preventing common errors like broken dependencies. It offers a CLI tool for local repository indexing and an MCP server to integrate this intelligence directly into AI-powered editors. Additionally, the platform features a web-based UI for visual exploration and enterprise-grade support for managed or self-hosted deployments.

Goose is a general-purpose AI agent designed to run locally on your machine for tasks ranging from coding and research to automation and data analysis. It is built in Rust to ensure high performance and portability across macOS, Linux, and Windows via desktop, CLI, and API interfaces. The project supports over 15 AI providers and integrates with more than 70 extensions through the Model Context Protocol.

k-skill is a collection of automation tools designed for AI agents to perform various tasks related to popular Korean services like SRT, KTX, Coupang, and KakaoTalk. It supports integration with major coding agents and allows users to execute tasks without needing additional client API layers. Users can get started by installing the full suite and configuring their credentials through the provided setup tool.

Paperclip is an open-source platform that provides a Node.js server and React UI to orchestrate teams of AI agents as a cohesive business entity. It functions like a task manager, offering features such as org charts, budget enforcement, and goal alignment to manage autonomous operations. Users can integrate their own agents to run businesses 24/7 while maintaining oversight through a centralized dashboard.

Onyx is a feature-rich open source AI platform designed to provide an easy-to-deploy application layer interface for large language models. The platform supports RAG, deep research, code execution, and various AI agent capabilities, while remaining compatible with mainstream self-hosted and proprietary LLMs. Users can deploy via the standard or lightweight versions to meet different needs ranging from personal use to enterprise-level collaboration.

OmniVoice is an advanced large-scale multilingual zero-shot speech synthesis model based on a diffusion language model architecture, supporting over 600 languages. The model features exceptional inference speed and enables high-quality voice cloning and voice design capabilities. Users can easily perform speech generation via Python API or command-line tools, with support for fine-grained non-linguistic symbols and pronunciation control.

AI Marketing Skills is an open-source project designed for marketing and sales teams, providing a series of ready-to-run automated workflows and scripts. These tools are designed to integrate with Claude Code or other AI coding agents, optimizing business processes through expert panels, scoring algorithms, and automated pipelines. The project covers professional skills in various fields, ranging from growth experiments and sales lead mining to content operations and financial analysis.