milu

What it does

Milu is a comprehensive AI agent framework designed for production environments. It enables building and deploying multi-user AI agents with support for 9 LLM providers, prioritizing Chinese language models like Qwen, DeepSeek, Kimi, and GLM. The framework abstracts away complexity of multi-agent coordination and LLM integration.

Key features

• Multi-LLM support with 9 providers and Chinese model prioritization
• Built-in tools, MCP (Model Context Protocol) integration, and RAG (Retrieval-Augmented Generation)
• Sub-agents and skills system for modular agent development
• Agent pool management and task scheduling capabilities
• Multi-tenant architecture for production deployments
• Flexible deployment: CLI, microservice, or embedded library

How to set up

Clone the repository from GitHub and install dependencies using pip or your preferred Python package manager. Configure your desired LLM provider credentials (Qwen, DeepSeek, Kimi, GLM, or others). Deploy as a standalone service, run via CLI with configuration files, or import as a Python library in your application. Comprehensive documentation includes examples for common use cases.