Packages

Demonstration project showing how to build SSE-based MCP servers and clients, enabling decoupled cloud-native communication between AI agents and tool providers. Includes a weather server using National Weather Service APIs as an example.

Bridges local LLMs running via Ollama to MCP servers, enabling open-source models (Qwen 2.5 7B recommended) to use MCP tools like filesystem, web search, GitHub, Gmail, and image generation through JSON-RPC protocol translation.

Chat with your documents offline using local AI models. Supports GGML/GGUF, Hugging Face Transformers, and GPTQ models with both web and CLI interfaces. Based on PrivateGPT with expanded format support.

Lightweight Python library for adding real-time multi-object tracking to any object detector with just a few lines of code. Uses customizable distance functions and Kalman filtering.

Educational repository with 30+ Jupyter notebooks covering AI agent frameworks (LangGraph, CrewAI, Autogen, OpenAI Swarm), multi-agent patterns, evaluation methods, and custom agent development for O'Reilly training courses.

A framework for rapidly building AI-native IDE products. Supports cloud, desktop (Electron), and web deployments with MCP client support for Model Context Protocol tools. Created by Alibaba/Ant Group.

Integration guide and community plugin for connecting Feishu (Lark) with OpenClaw AI agents, enabling document creation, calendar management, task handling, and messaging through Feishu bots.

Open source LMS REST API exposing 700+ web service functions for managing courses, users, grades, assignments, and learning content across the world's most widely deployed e-learning platform.

Cross-platform desktop AI assistant and MCP client that integrates multiple AI providers (OpenAI, Anthropic, Google, etc.) with local knowledge base support and MCP server tool access.

Intelligent LLM router that directs requests to optimal models across 41+ models and 7 providers based on cost and capability, using 15-dimension scoring with sub-millisecond routing and USDC crypto payments.

A Docker image combining Uvicorn and Gunicorn for running Python ASGI web applications with automatic worker scaling based on CPU cores. Now deprecated in favor of using plain Uvicorn with --workers flag.

LinkedIn's REST API for reading profile data, posting content to LinkedIn feeds, managing company pages, and accessing marketing/recruiting tools for approved partners.

Python library providing a Tkinter table widget for interactive data analysis using pandas DataFrames, with built-in plotting via matplotlib. Includes a standalone DataExplore application for spreadsheet-like data exploration.

Desktop companion application for managing OpenClaw AI agents. Provides a visual interface for agent configuration, model profile management, Discord channel bindings, diagnostics, and version control with snapshot history and rollback — replacing manual JSON editing.

Official TanStack CLI for creating and managing TanStack Router and Start applications, with project scaffolding, add-on integrations (auth, database, deployment), and agent-friendly JSON output for discovery.

A desktop MCP client application that provides a unified interface for interacting with multiple LLM providers (ChatGPT, Claude, Qwen, etc.) while supporting MCP tool integration, agent configuration, and visual tool call tracing.

Open tools and data for building cloudless automatic speech recognition models using Kaldi, KenLM, and wav2letter++.

A curated list of 28+ OpenClaw-inspired AI agent implementations across multiple languages (TypeScript, Go, Rust, Python, C, Zig), focusing on lightweight, self-hosted, privacy-aware agent architectures.

API proxy for Midjourney image generation that wraps Discord-based interactions into a REST API. Supports imagine, blend, describe, shorten commands with multi-account load balancing, real-time task progress, Chinese prompt translation, and face-swap capabilities.

Open-source agent system suite from Kuaishou Technology providing agent frameworks, fine-tuned language models (KAgentLMs), 200K+ training examples, and a 3K+ item benchmark for evaluating agent planning, tool-use, reflection, and reasoning capabilities.