Starred repositories
JiuwenSwarm is an intelligent AI Agent built on openJiuwen. It extends the powerful capabilities of large language models directly to your fingertips through various communication apps you use daily.
Script to build possibly the most trimmed-down Windows 11 experience.
A high-throughput and memory-efficient inference and serving engine for LLMs
A lightweight, lightning-fast, in-process vector database
Open-source meeting transcription API for Google Meet, Microsoft Teams & Zoom. Auto-join bots, real-time WebSocket transcripts, MCP server for AI agents. Self-host or use hosted SaaS.
Harness engineering beginner tutorial, from 0 to 1
Build distributed, production-grade, long-running agents.
分享一些好用的 Dify DSL 工作流程，自用、学习两相宜。 Sharing some Dify workflows.
OpenAI-compatible proxy that stacks the free tiers of 28 LLM providers (~4B tokens/month) behind one /v1 endpoint — plus any custom OpenAI-compatible endpoint. Smart routing, automatic failover, en…
📑 PageIndex: Document Index for Vectorless, Reasoning-based RAG
The agent that grows with you
A complete AI agency at your fingertips - From frontend wizards to Reddit community ninjas, from whimsy injectors to reality checkers. Each agent is a specialized expert with personality, processes…
AI turns documents or topics into real, native PowerPoint decks—with native shapes, transitions and animations, data-backed charts and tables on demand, audio narration from speaker notes, and supp…
2026.3.31 claude code 意外把包含源码的文件上传到 npm 仓库，版本号是 2.1.88，其中 cli.js.map 文件有 57MB 的体积，claude code 的源码在该文件的 sourcesContent 字段里面，解压还原后有 70w 行代码
Proxifier Alternative to redirect any Windows/MacOS/Linux TCP and UDP traffic to HTTP/Socks5 proxy
Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞
A curated list of awesome Claude Skills, resources, and tools for customizing Claude AI workflows
An Application Framework for AI Engineering
📚 《从零开始构建智能体》——从零开始的智能体原理与实践教程
慢sql语句检测平台 capture and explain sqls, then you know whether it's slow or fast.
Agent framework for the JVM. Pronounced Em-BAY-bel /ɛmˈbeɪbəl/
"RAG-Anything: All-in-One RAG Framework"
A free, open source, and extensible speech-to-text application that works completely offline.
An Engine-Agnostic Deep Learning Framework in Java