28 projects Original English introductions · click a card to open the repo. Sort by stars or last update; data refreshes weekly.
The "build it yourself" approach to learning: recreate classic technologies — databases, operating systems, compilers, search engines — from scratch. One of the most-starred repos on GitHub.
A free programming books encyclopedia: an index of thousands of free ebooks and courses in hundreds of languages — a treasure trove for self-learners and developers.
A personal AI assistant that runs entirely on your device: connects 50+ apps (WeChat, WhatsApp, Slack, Telegram…), browses and fills in forms, runs commands, controls your smart home, and can even teach itself new skills. The fastest-growing open-source project of 2026.
A "superpowers" skill pack for AI coding agents: debugging playbooks, planning flows, testing strategies and more — so agents genuinely work the way experienced engineers do.
A performance-optimization system for AI agents: gives coding agents like Claude Code and Codex skills, instincts, memory and safety guardrails — like installing an operating system on your agent.
A huge collection of the Claude Code ecosystem: skills, plugins, configuration and workflows all in one place. Essential reading if you want to use Claude Code deeply.
A visual, drag-and-drop workflow automation platform with native AI nodes and LangChain integration, 400+ service connectors, and self-hostable deployment — the go-to choice for enterprise automation.
The pioneer of AI agents: a platform that plans and executes multi-step tasks autonomously, with plugin extensions, web search and code execution. Fully rewritten as a visual agent builder.
Run open-source LLMs like Llama, DeepSeek and Gemma locally with a single command. Written in Go, lightweight and efficient — a core component of local AI infrastructure.
An official Microsoft tool that converts anything — Word, PDF, Excel, PPT, audio — into Markdown. The universal pre-processing step before feeding content to AI.
A production-ready AI application development platform: built-in RAG pipeline, multi-model support, agent orchestration, usage monitoring and native MCP support. Built by a Chinese team.
A visual agent / RAG building platform based on LangChain: drag and drop to assemble multi-agent flows, deploy as an API in one click. Acquired by DataStax.
A feature-complete, self-hosted AI chat platform: connect local or cloud models, with built-in RAG, voice, and multi-user permissions. Spin up your own private ChatGPT with one command.
The connective tissue of LLM application development: a one-stop component library for chains, agents, memory, retrieval and tool calling — the ecosystem standard supported by every major model vendor.
A cross-platform desktop tool to manage configuration switching for Claude Code, Codex, OpenCode and Gemini CLI from a single panel. A must-have for heavy multi-agent users.
Turn any codebase — along with its docs, SQL and config — into a knowledge graph so AI agents can truly "understand" the whole project instead of just reading files.
A real-time global intelligence dashboard: AI aggregates news, social media and market signals into a visual command center for intelligence workflows.
An adaptive web-scraping framework: anti-bot bypass, self-healing selectors, and everything from single-page requests to full-site crawls. A new benchmark in scraper engineering.
Let AI agents "see" the entire internet: a unified API to read and search content from Twitter, Reddit, YouTube, GitHub and more.
A context-optimization layer for LLM applications: automatically compresses and manages the context window. A lifesaver for long conversations and large codebases.
A drag-and-drop builder designed specifically for LLM applications, deeply integrated with LangChain. Supports Chatflow and Agentflow modes, with one-click Docker launch.
An enhanced ChatGPT clone: aggregates multiple model providers, supports Agents and MCP, multi-user account management, file uploads and a plugin system. Self-hostable.
A persistent knowledge graph of your codebase for MCP: 35 languages, millisecond queries, up to 99% faster than grep. Gives coding agents long-term memory.
An official Microsoft MCP server for browser automation: lets AI agents truly control a browser — click, fill forms, screenshot and scrape data fully automatically.
The official GitHub MCP server: lets AI agents manage repositories, PRs and issues directly — a core component for automation workflows.
An AI coding agent that lives in your terminal: hash-anchored context management keeps long sessions from drifting. A handy tool for heavy CLI users.
Command Blender with natural language: modeling, scenes and materials all done conversationally. The flagship project for AI-powered 3D.
A 20MB cross-platform database client with built-in MCP support — a lightweight way to let AI query your databases directly.
By Stars
By Last Update