AgentScope 1.0: A Developer-Centric Framework for Building Agentic Applications
Dawei Gao, Zitao Li, Yuexiang Xie, Weirui Kuang, Liuyi Yao, Bingchen Qian, Zhijian Ma, Yue Cui, Haohao Luo, Shen Li, Lu Yi, Yi Yu, Shiqi He, Zhiling Luo, Wenmeng Zhou, Zhicheng Zhang, Xuguang He, Ziqian Chen, Weikai Liao, Farruh Isakulovich Kushnazarov, Yaliang Li, Bolin Ding
2025-08-25
Summary
This paper introduces AgentScope, a new system designed to make it easier to build and manage complex 'agents' powered by large language models. These agents can use tools and interact with their environment to complete tasks, and AgentScope aims to provide all the necessary building blocks for developers.
What's the problem?
Building agents that can reliably use tools and interact with the real world is really hard. Existing systems are often complicated, don't easily adapt to new models or tools, and make it difficult to understand what the agent is doing or ensure it's safe. Developers need a more streamlined and robust way to create these kinds of applications.
What's the solution?
AgentScope solves this by providing a set of core components and standardized ways for these components to work together. It's built around a well-known approach called 'ReAct' which helps agents think through problems and take actions. The system is designed to handle many things happening at once, making it faster and allowing for more complex interactions between agents and people. It also includes tools for testing, visualizing what the agent is doing, and safely running the agent in a real-world setting.
Why it matters?
AgentScope is important because it lowers the barrier to entry for building powerful, AI-driven agents. By providing a solid foundation and helpful tools, it allows developers to focus on creating innovative applications instead of struggling with the underlying technical complexities. This could lead to more widespread use of agents in various fields, automating tasks and solving problems in new ways.
Abstract
Driven by rapid advancements of Large Language Models (LLMs), agents are empowered to combine intrinsic knowledge with dynamic tool use, greatly enhancing their capacity to address real-world tasks. In line with such an evolution, AgentScope introduces major improvements in a new version (1.0), towards comprehensively supporting flexible and efficient tool-based agent-environment interactions for building agentic applications. Specifically, we abstract foundational components essential for agentic applications and provide unified interfaces and extensible modules, enabling developers to easily leverage the latest progress, such as new models and MCPs. Furthermore, we ground agent behaviors in the ReAct paradigm and offer advanced agent-level infrastructure based on a systematic asynchronous design, which enriches both human-agent and agent-agent interaction patterns while improving execution efficiency. Building on this foundation, we integrate several built-in agents tailored to specific practical scenarios. AgentScope also includes robust engineering support for developer-friendly experiences. We provide a scalable evaluation module with a visual studio interface, making the development of long-trajectory agentic applications more manageable and easier to trace. In addition, AgentScope offers a runtime sandbox to ensure safe agent execution and facilitates rapid deployment in production environments. With these enhancements, AgentScope provides a practical foundation for building scalable, adaptive, and effective agentic applications.