把最丝滑的通用 Agent 体验带进你的工作流，为 100x 专业用户而生的未来产品，正在实现 proactive Agent 阶段。基于 Claude Agent SDK 的完整开源实践，原生支持飞书群聊调用、灵活接入任意大模型供应商 —— 让顶级 Agent 能力真正跑在你每天用的地方。
Tech Interview Handbook is a free and curated collection of technical interview preparation materials for busy engineers, authored by Yangshun (creator of Blind 75). It goes beyond typical algorithm questions, covering essential aspects like resume building, behavioral interview questions, and front-end specific interviews. The handbook aims to provide practical, condensed content to help job seekers efficiently prepare for and succeed in technical interviews, offering a comprehensive guide beyond just coding challenges.
OpenSpec, by Fission-AI, is a lightweight specification framework designed to bring predictability to AI-assisted coding. It introduces an artifact-guided workflow, enabling human and AI to align on proposals, specs, design, and tasks before code implementation. This ensures clarity and organization, supports iterative development, and integrates with over 20 existing AI assistants.
🤖 A client-first, server-capable, fully type-safe router and full-stack framework for the web (React and more).
Maxun is a powerful, open-source, no-code platform designed to transform any website into structured, reliable data or a real-time API. It unifies essential web data operations including real-time scraping, intelligent crawling, automated search, and advanced AI-powered data extraction. Users can leverage intuitive Recorder Mode to capture actions or employ LLM-driven AI Mode for natural language data extraction. Maxun offers solutions for converting full webpages into clean Markdown/HTML, managing complex pagination, and scheduling automated workflows. Ideal for lead generation, market research, and content aggregation, it empowers users to efficiently collect and organize web data without writing a single line of code, providing self-hostable flexibility and robust developer tools.
EmDash is a modern, full-stack TypeScript Content Management System built on Astro and Cloudflare. It re-imagines the extensibility and user experience of traditional CMS platforms like WordPress, leveraging a serverless, type-safe architecture. A core innovation is its sandboxed plugin ecosystem, where plugins run in isolated Cloudflare Worker sandboxes with explicit capability declarations, fundamentally addressing security vulnerabilities. EmDash utilizes Portable Text for structured JSON content storage, decoupling content from presentation and enabling flexible multi-channel distribution. Designed for AI Agents, it offers development skill files, programmatic CLI tools, and a built-in MCP server for direct AI interaction. EmDash supports various databases and S3-compatible storage, providing a high-performance, secure, and developer-friendly platform for modern web content management.
Personal AI Infrastructure (PAI) is a comprehensive "Life Operating System" designed by Daniel Miessler. Unlike standard agent harnesses, PAI serves as an identity and goal-oriented layer that helps users transition from their current state to an "Ideal State" (ISA) using Algorithm v6.3.0. It features the Pulse dashboard and a Digital Assistant (DA) interface. Built on a "text-over-opaque-storage" philosophy, it prioritizes filesystem context over RAG and uses deterministic code-based skills to automate and magnify human capabilities.
Firecrawl is the Web Data API for AI, empowering AI agents with clean, reliable web data. It provides robust capabilities to search, scrape, and interact with the web at scale, boasting industry-leading reliability that covers 96% of pages, including JS-heavy content. Delivering LLM-ready outputs like clean Markdown, structured JSON, and screenshots, Firecrawl significantly streamlines data acquisition for AI applications. It handles complex web scraping challenges such as proxy management, rate limits, and JavaScript-blocked content, allowing developers to focus on building intelligent agents and dynamic apps. Seamlessly integrate Firecrawl to enable real-time, precise web access and interaction.
Flowise is a leading open-source visual AI Agent builder designed to significantly simplify the development and deployment of Large Language Model (LLM) applications. It empowers users to effortlessly design, assemble, and deploy complex AI agents through an intuitive drag-and-drop interface, enabling sophisticated logic flows without extensive coding. Key advantages include its exceptional customizability, modular architecture, and flexible deployment options, supporting quick setup via Docker, npm, or self-hosting on various cloud platforms (AWS, Azure, GCP, Alibaba Cloud). Flowise accelerates the realization of AI solutions, making it an ideal choice for individual developers and enterprise teams to build, test, and manage AI Agents efficiently.
The "Maths, CS & AI Compendium" is an open, unconventional textbook designed for curious practitioners seeking a deep understanding of mathematics, computing, and artificial intelligence from the ground up, rather than just passing exams or interviews. It prioritizes intuition, real-world context, and clear explanations, avoiding dense notation. The project also includes an MCP Server, allowing AI assistants like Claude Code, Cursor, and VS Code to utilize the compendium as a knowledge base for educational purposes and example implementations. It covers a broad range of topics from fundamental concepts to advanced areas such as machine learning and AI inference.
VoltAgent is an end-to-end AI Agent Engineering Platform empowering developers with full code control and production-grade operational visibility. It comprises a powerful open-source TypeScript framework and the VoltOps Console. The framework supports building intelligent, multi-step, multi-agent systems with comprehensive features including memory management, RAG integration, guardrails, declarative workflow orchestration, and a versatile tool registry, ensuring compatibility with major LLM providers. The VoltOps Console (cloud/self-hosted) delivers essential observability, automation, deployment, evaluation, and prompt management. This integrated platform enables developers to efficiently build, deploy, and monitor complex AI Agents from concept to production.
OpenSpec by Fission-AI is a lightweight specification framework designed to address the unpredictable output of AI coding assistants when requirements reside only in chat history. It introduces an artifact-guided workflow that enables human and AI alignment on specifications before any code is written. Users initiate an iterative process with commands like `/opsx:propose`, creating dedicated folders for each change encompassing proposals, specs, design, and task lists, ensuring fluid, organized, and scalable development. It integrates with over 20 AI assistants and various development tools, suitable for projects ranging from personal to enterprise-level.
Desktop CC GUI is an open-source, next-generation VibeCoding editor designed for professional developers, serving as an alternative to Cursor with a focus on developer experience. It integrates multiple AI coding engines like Claude Code and Codex, offers professional development panels (chat canvas, built-in terminal, Git panel, Kanban, planning), an AI memory system (project memory, skills, prompt library), and MCP protocol support. It provides a cross-platform native experience on macOS, Windows, and Linux, emphasizing transparency and development efficiency.
Claude HUD is a plugin designed for Claude Code, providing real-time session insights. It integrates with Claude Code's native statusline API to continuously display critical information below the input area, including context window usage, active tools, running agents and their progress, and todo completion. Leveraging native token data, it dynamically scales with the context window size, supports multi-language display, and offers extensive customization, helping users effectively monitor and manage their coding sessions.
AI Novel Writing Assistant is an AI-native open-source project designed for long-form novel creation. Moving beyond simple sentence completion, it acts as a comprehensive AI production system that guides users from a vague idea to a complete novel. It automatically constructs worldviews, characters, and plot structures, manages knowledge and settings via RAG, and maintains writing style and narrative consistency. Targeting beginner writers, the system leverages AI as a core orchestrator for planning, scheduling, and execution, focusing on completing entire novels. It is available as a Windows desktop application and an open-source monorepo for AI-native product and agent workflow research.
Context7, developed by Upstash, is a platform designed to provide Large Language Models (LLMs) with up-to-date, version-specific code documentation. It solves the problem of LLMs generating outdated, hallucinated, or generic code examples by injecting real-time, source-driven documentation and code examples directly into the LLM's context. Context7 supports both CLI + Skills and MCP (Multi-Client Protocol) modes, ensuring that developers' AI coding assistants always access accurate and current information, thereby improving the quality and reliability of code generation.
Archon is the first open-source harness builder and workflow engine for AI coding, designed to make AI coding deterministic and repeatable. It allows developers to define their entire development processes—including planning, implementation, validation, code review, and PR creation—as YAML workflows. This ensures reliable and consistent execution across projects, preventing the randomness often seen in AI agent outputs. By encoding the development process with defined phases and validation gates, Archon provides a structured framework where AI intelligently fills in each step, akin to how Dockerfiles brought determinism to infrastructure or GitHub Actions to CI/CD.
VibeClaw, an innovative browser sandbox version of the OpenClaw project, is engineered to significantly lower the entry barrier for AI Agent adoption. It provides a lightweight, instant browser-based environment, enabling users to launch and experience powerful AI Agents within just one second, without any installation. This core feature eliminates the complexities of traditional software setup, allowing beginners, developers, and presenters alike to rapidly access AI Agent capabilities within the OpenClaw ecosystem at zero cost and zero waiting time. VibeClaw dramatically enhances the universality and accessibility of AI Agent technology, making it ideal for rapid prototyping, product demonstrations, educational experiences, and initial feature exploration.
OpenClaw is a leading open-source autonomous AI Agent platform, empowering individuals to deploy and run personalized AI assistants directly on their local devices. It seamlessly connects large language models with local file systems via integration with over 20 mainstream messaging applications like WhatsApp, Telegram, and Discord. The platform offers robust multimodal interaction capabilities, including Shell command execution, file management, web automation, voice dictation, and real-time canvas control. Designed for a fast, always-on, and highly customizable personal assistant experience, OpenClaw prioritizes user privacy through strict DM security policies. With over 336K GitHub stars, it stands as one of the largest AI Agent open-source projects, driving device intelligence and task automation.
chrome-devtools-mcp empowers AI coding agents such as Gemini, Claude, Cursor, or Copilot to fully control and inspect a live Chrome browser. Acting as a Model-Context-Protocol (MCP) server, it grants your AI assistant access to the comprehensive capabilities of Chrome DevTools. This tool facilitates reliable browser automation, in-depth debugging, and robust performance analysis, enabling AI agents to interact with web environments more intelligently. It streamlines development workflows and assists in resolving complex web-related challenges by leveraging the full power of Chrome's native developer tools.
NanoClaw is a security-focused, containerized AI Agent solution built with the Anthropic Agent SDK, recognized as the industry's "Security Champion." As an enhanced version of OpenClaw, it was entirely rewritten in TypeScript, with its core code streamlined to approximately 500 lines to achieve an ultra-minimal attack surface. This design significantly reduces potential vulnerabilities and enhances auditability and transparency. NanoClaw provides a highly secure, easily deployable, and reliable AI Agent environment, ideally suited for critical applications in finance, healthcare, and government sectors with stringent data security and compliance requirements.
Page Agent, developed by Alibaba, is an innovative JavaScript-based in-page GUI agent designed to precisely control web interfaces using natural language commands. Operating entirely within the webpage, it employs text-based DOM manipulation, eliminating the need for browser extensions, Python, headless browsers, or multi-modal LLMs and screenshots. This ensures exceptional ease of integration and high efficiency. Users can seamlessly integrate their own Large Language Models. Key applications include rapidly deploying SaaS AI copilots, automating complex enterprise form filling (e.g., ERP/CRM), enhancing web accessibility, and supporting multi-page agent tasks with an optional Chrome extension. Page Agent transforms complex workflows into simple natural language commands, significantly boosting productivity and user experience.
OpenClaude, developed by Gitlawb, is an open-source command-line interface (CLI) coding agent designed to provide developers with a unified and powerful AI-assisted programming environment. It supports a vast array of cloud and local large language models, including OpenAI, Gemini, GitHub Models, DeepSeek, Ollama, and connects to over 200 models via OpenAI-compatible APIs. OpenClaude's core strength lies in its terminal-first workflow, integrating rich tool-driven coding capabilities like Bash execution, file I/O, grep, glob, and multi-step tool calling. It also features streaming responses, image input, configurable agent routing, and web search/fetching. With its robust agent capabilities, OpenClaude seamlessly integrates into CI/CD pipelines or custom applications, serving as an ideal bridge between diverse AI models and practical development workflows.
AionUi, developed by iOfficeAI, is a free, open-source, cross-platform desktop application that unifies various command-line AI tools (such as Gemini CLI, Claude Code, and others) into a single, intuitive graphical user interface. It functions as a "Cowork" platform where AI agents operate alongside users on their computers, facilitating autonomous task execution, file operations, web browsing, and 24/7 automation. AionUi features multi-session chat with independent context, local data storage, comprehensive multi-model support (including local models), multi-agent capabilities, smart file management, AI image generation, and remote access via WebUI and popular messaging applications.
Pi Monorepo by badlogic is an integrated toolkit that serves as a backbone for the OpenClaw ecosystem, focused on building AI agents and managing Large Language Model (LLM) deployments. It functions as a minimalist terminal coding harness, providing a unified multi-provider LLM API, an agent runtime with tool calling and state management, an interactive coding agent CLI, a Slack bot, and a CLI for managing vLLM deployments on GPU pods. The project emphasizes its extensibility, allowing users to customize workflows with TypeScript extensions, skills, prompt templates, and themes, and encourages sharing open-source coding agent session data to continuously improve agent performance.
Dexter is an autonomous financial research agent that thinks, plans, and learns. It decomposes complex queries into structured tasks, executes them using real-time market data, and employs self-reflection to validate results. Built with Bun and supporting multiple LLMs, it provides data-backed financial insights with integrated safety features.
n8n is a secure workflow automation platform for technical teams, uniquely combining the rapid development of no-code with the powerful flexibility of custom code. It enables users to build sophisticated automations via a visual interface, seamlessly integrating JavaScript/Python code and npm packages. A core strength is its native AI capabilities, particularly for constructing LangChain-based AI Agent workflows utilizing proprietary data and models. With over 400 integrations and 900 ready-to-use templates, n8n offers extensive connectivity. Operating under a fair-code license, it provides full control over data and deployments, whether self-hosted or through its cloud service, complemented by enterprise features like advanced permissions and SSO.
wechat-article-exporter is an open-source tool designed for batch downloading WeChat Official Account articles. By leveraging the internal search API of the WeChat backend, it allows users to export articles into formats like HTML, Markdown, and Excel. It features 100% layout restoration for HTML files and supports fetching metadata such as read counts and comments. It can be accessed via an online web service or self-hosted using Docker or Cloudflare Workers for private data collection and archiving.
12-factor-agents is a methodology developed by humanlayer for building reliable, production-grade LLM applications. Inspired by the original 12-Factor App, it provides 12 core engineering principles to make AI agents more predictable and scalable. Key concepts include owning your prompt and context window, unifying execution state, using tools as structured outputs, and designing agents as stateless reducers. It shifts the paradigm from autonomous 'black-box' loops to robust software engineering practices for AI.
Prompt Optimizer by linshenkx is a powerful AI prompt optimization tool designed to enhance the quality of AI model outputs. It intelligently optimizes both system and user prompts through multi-round iterative improvements, supporting analysis, evaluation, and comparison of different optimization results. The tool integrates mainstream AI models like OpenAI and Gemini, offers text and image generation (T2I/I2I), and includes an advanced testing mode with context variable management, multi-turn conversations, and function calling. Featuring a secure client-side architecture, it supports web, desktop, Chrome extension, and Docker deployments, and is compatible with the MCP Protocol.