agentscope

AgentScope is a framework designed to simplify the creation of LLM-powered multi-agent applications. It offers tools, documentation, and a drag-and-drop workstation for building and deploying agent-based systems. AgentScope is a

7,239
394

AgentScope: README Summary

AgentScope is a platform designed to simplify the development of multi-agent applications powered by large language models (LLMs). It offers an easy-to-use interface, high robustness, and actor-based distribution for streamlined development.

Key features include fruitful components, comprehensive documentation, and broad compatibility. The AgentScope Workstation provides a drag-and-drop programming platform and a copilot for beginners. It supports customized fault-tolerance controls and retry mechanisms to enhance application stability.

AgentScope supports various model libraries, including OpenAI, DashScope, Gemini, ZhipuAI, ollama, LiteLLM, Yi API, and Anthropic API, through ModelWrapper. It also supports local model deployment using libraries like ollama, Flask, Transformers, ModelScope, FastChat, and vllm.

Example applications include basic and self-organizing conversations, integration with LangChain, ReAct agents, natural language to SQL queries, RAG agents, and games like Gomoku and Werewolf. AgentScope is released under the Apache License 2.0, and contributions are welcome.

Repository

MO
modelscope

modelscope/agentscope

Created

January 12, 2024

Updated

March 28, 2025

Language

Python

Category

AI